|
|
|
|
 Home
Page
Search
Final
Report
Interim
Report
Evidence
- Transcripts
- Witness Statements
- Submissions
- CCNR Report and associated
documents
- Statistical Reports and associated
documents
- Background Papers
- Expert Papers
Inquiry
Seminars
About
the Inquiry
Help
|
 |
Daily transcripts
Hearing summary
14 July 1999
Today the Inquiry hearings concluded a two-day look at data sources and
the relevance of statistical evidence to the Inquiry’s terms of reference. Mr Andrew
Hooper, Health Records Manager, Bristol Royal Infirmary (BRI) and United Bristol
Healthcare NHS Trust (UBHT); Ms Ann Harding, Acting Director, NHS Information Authority;
and Dr David Spiegelhalter, Senior Scientist, Institute of Public Health, and a member of
the Inquiry’s expert group of witnesses, discussed the introduction, quality and
consistency of the data (administrative not clinical) recorded on the Patient
Administration System (PAS) at the BRI. Ms Harding then went on to clarify some points
arising from Mr Willmer’s evidence given yesterday relating to remuneration and
training for clinical coders and incentives to promote accuracy of coding, which she said
relied largely on personal motivation. Mr Spiegelhalter then reviewed the past two days,
commenting on the usefulness to the Inquiry of the data sources discussed and concluded by
explaining the next stages in the process of analysing the data.
The hearings were adjourned until Monday 19 July.
Full copies of the documents "Preliminary Critical Overview
of Existing Data Sources Relevant to the Inquiry Remit" and "The Inquiry’s
Approach to Making Use of Relevant Data Sources" are published the Inquiry
website under Inquiry Procedures.
|
FULL TRANSCRIPT
1 Day 39, 14th July 1999
2 (10.00 am).
3 MR LANGSTAFF: Good morning, sir. Today the second part of
4 the preliminary overview, preliminary stage, in
5 examining the quality of the data before we approach the
6 analyses and draw the lessons that can be learned from
7 them, if any. We will be looking at the PAS (Patient
8 Administration System) which the hospitals in Bristol,
9 in particular the Trust, had during the years which are
10 relevant to us.
11 This morning we will hear first of all from
12 Mr Andrew Hooper; we will then hear from
13 Miss Ann Harding, our expert, who will pick up some of
14 the points which Mr Willmer made yesterday. She, of
15 course, while Mr Hooper is giving his evidence, will,
16 I hope, interact with him in the same way as we achieved
17 yesterday when we heard the evidence of Mr Willmer in
18 the afternoon.
19 Then Dr Spiegelhalter will finish the session with
20 examining the way forward that the Inquiry can take.
21 We begin with Mr Andrew Hooper. Will you come
22 forward, please?
23 Mr Hooper, would you mind standing for the oath?
24 MR ANDREW HOOPER (AFFIRMED):
25 Examined by MR LANGSTAFF:
0001
1 Q. Can we have on screen WIT 211/1? Is that the first
2 page of a statement which you made for the purposes of
3 this Inquiry about the Patient Administration System in
4 operation at the Bristol Royal Infirmary?
5 A. Yes.
6 Q. If we go to page 7, that is your signature?
7 A. Yes.
8 Q. And the contents of that statement are true and
9 accurate, are they?
10 A. Yes.
11 Q. You also have provided for us -- we find this at
12 INQ 1/24 -- comments on the published preliminary
13 overview of existing data sources, the paper produced by
14 the Inquiry Secretariat and published on the Internet,
15 and dated 13th July. It is one page and we can see if
16 we scroll down your signature at the bottom.
17 A. Yes.
18 Q. You have also provided us with a number of
19 appendices which show various diagrams and attach
20 a number of letters in relation to the Patient
21 Administration System.
22 A. Yes.
23 Q. Can I go back, then, to explore the nature of the
24 Patient Administration System? What we will do is, we
25 will take your statement as read and therefore I shall
0002
1 not ask you to go over everything which is in it and
2 just explore one or two aspects of it or that arise from
3 it.
4 You make the point that you have worked in
5 health records initially as a records clerk and then as
6 a records manager, and finally as the Health Records
7 Manager for the Trust over a number of years, and
8 certainly throughout the years with which this Inquiry
9 is concerned.
10 Since 1988 you have worked in the Bristol
11 hospitals?
12 A. Yes, I was based at the Trust in 1988, based at the
13 Bristol Royal Infirmary but with Trust responsibilities.
14 Q. You told us that in April 1988, when you became the
15 Trust PAS manager, you began to introduce a PAS system
16 to the Trust, having had familiarity with it earlier in
17 your post in Southmead?
18 A. Yes.
19 Q. Am I right in thinking that the inpatient part of
20 the PAS system was operative from 1st April 1988?
21 A. What I actually did, when I came back to Bristol
22 & Weston Health Authority, I had a three and four year
23 plan at that stage which was to implement a master
24 patient index, an inpatient module, an outpatient module
25 and a waiting list module. That was my remit when I was
0003
1 re-employed back there in 1988. The first thing I did
2 was to set up the index and because we did not want to
3 be in a situation where we had a manual and a computer
4 system, we actually took on all the retrospective, all
5 the historical index right back until 1948.
6 Historically, the patients' demographic information had
7 always been kept on 5 x 3 index cards so the first part
8 of the process was to convert the old manual system
9 across to the new Patient Administration System, which
10 at that time was a system being supported by the South
11 West Regional Health Authority.
12 So that was the first task, to achieve that index.
13 Q. Can I ask you to pause there for a moment? The reason
14 for that is that you are speaking naturally quite fast.
15 We have stenographers who need to keep up with you.
16 They are taking a note of every word you speak. If you
17 do not mind either speaking a little more slowly, or if
18 that is unnatural, I will ask you to pause for a few
19 moments.
20 I stopped you in mid-flow.
21 A. The first part was to set up the index. Once that had
22 been achieved, there was a local system that had been
23 developed in-house at Bristol at that stage. In
24 essence, all that was, it was a way of forming a very
25 brief index of the patients that had attended. They
0004
1 used to be called data admission and data discharge, and
2 they also used it as a diary mechanism for outpatients.
3 They used to use it for producing clinic lists and
4 patients' labels, no more, no less. It had no links to
5 information systems at all at that time.
6 Because they had that system in place, when I set
7 up the inpatient module on the new system, which would
8 have been some time probably in late 1989, we loaded
9 back retrospectively the inpatient episodal information
10 to 1st April 1988 and some of that information was
11 lifted from the locally developed system -- not very
12 much -- and then from a date which I cannot remember, as
13 I say, it was in 1989, we started entering real-time all
14 the admission, the transfer and the discharge data.
15 Q. So from 1st April 1988 up until a date in 1989, the data
16 is derived from historical records and loaded into the
17 computerised PAS system?
18 A. Yes.
19 Q. And from that time onwards, the information is recorded
20 in real-time?
21 A. Yes. What we actually did, we got rid of the old manual
22 bed statements and replaced it with real-time entry of
23 the data. That was the carrot we dropped in front of
24 the nurses and the ward staff to get the information put
25 on to the system. We wanted to try and encourage them
0005
1 to put in the data real-time.
2 Q. Did it work?
3 A. Yes.
4 Q. So effectively, if we look at it as a system -- first of
5 all, I want to explore with you, really, what it
6 recorded. Its purpose was administrative as you point
7 out. You point that out more than once in answer or
8 response to the invitation to comment on the critical
9 overview that was published?
10 A. Yes.
11 Q. Although it was administrative, it would record the
12 demographic information about the patient, date of
13 birth, postcode, date of admission, if admitted?
14 A. Yes.
15 Q. Date of outpatient attendance. You deal with
16 outpatients as well as inpatients, presumably?
17 A. Because it was supported by the regional computer
18 centre, as it was then, the system has always complied
19 with mandatory requirements so any mandatory information
20 that was required locally or nationally, it could always
21 be collected through the system. So, for instance, for
22 inpatients, it would collect all the patient's
23 demographic information, name, date of birth, address,
24 GP, and you would go on through that process to put on
25 the information required related to the date of
0006
1 admission, where they were referred from and so on and
2 so forth, the referring person, and then any internal
3 movements within that episode, whether it would be the
4 consultant transfer or a ward transfer, that was also
5 put on to the system in real-time, and then obviously
6 once the patient was discharged home, that should be
7 done the moment the person walks out of the door or is
8 taken out of the ward.
9 So the system has always complied with national
10 statistical requirements.
11 Q. And it would include, would it, a description of the
12 diagnosis?
13 A. No, not at all. Not at that stage. The diagnostic
14 information would have been put on after the patient had
15 been discharged through a diagnostic coding routine that
16 was linked in to the inpatient module.
17 Q. So if one looks at the PAD, the patient administration
18 database, one will find, will one, the diagnostic
19 information, even though it was not put on in real-time
20 as you described?
21 A. Yes. Once the diagnostic information has been input on
22 to the PAS, it would then move across to any other
23 systems that required that information.
24 Q. So if one goes back to the PAS now, one will find on
25 it diagnostic information and information as to any
0007
1 operative procedure which was carried out?
2 A. Back to 1st April 1988, yes.
3 Q. The system will record, will it, the death, if death
4 has happened, of the patient?
5 A. If the patient dies in hospital, then obviously, when
6 they are discharged from the ward, the disposal is put
7 on as "deceased". That will immediately update the rest
8 of the system and any outstanding outpatient
9 appointments, waiting list entries, will automatically
10 be cancelled off the system.
11 Similarly, when the patient's discharge summary is
12 produced, it should say "deceased" on the discharge
13 summary. That information will then go back to the GP.
14 There will obviously be the death notification as well.
15 Where we do have a bit of a problem is sometimes getting
16 the information back for the patients that have died
17 outside of the hospital. The information we get is only
18 as good as the information given to us by the GP's
19 surgery or the relatives.
20 Q. One of the differences, I think, that we have been told
21 of, I have been told of, comment on it if you can,
22 between HES and the information available from the PAS
23 system, is that HES takes, as it were, a snapshot
24 because it asks for data at intervals from regions and
25 hospitals, and so information that it has as to death
0008
1 will be the status as then revealed by the PAS records.
2 But the database which you have, the Patient
3 Administration System will be dynamic in the sense that
4 it will record deaths whenever they are notified to the
5 hospital, presumably?
6 A. That is correct, yes.
7 Q. So one will expect the PAS system to show greater
8 mortality rates than the HES system?
9 A. I would agree with that statement, yes.
10 Q. Can I, perhaps with the help of Ann Harding, or
11 Dr Spiegelhalter, explore the accuracy of the death
12 information that gets on to your system here in Bristol
13 and indeed for that matter, any comparative system
14 elsewhere.
15 Ann, would you wish to ask --
16 MISS HARDING: I think there are a couple of important
17 issues here. A death associated with a particular
18 episode of care and a death that is associated only as
19 the subject of a record as part of the administrative
20 system: I just wonder if you could comment how, in your
21 PAS, you handled that, whether it was possible to handle
22 that within the PAS system or whether it was not
23 possible to handle it within the PAS system?
24 A. I am sorry, could you just repeat the question? I am
25 not sure of the first bit.
0009
1 Q. A death associated with a particular episode of care
2 which would be recorded on the HES database associated
3 with that episode: were you able, within your Patient
4 Administration System, to associate a death that did not
5 happen immediately with an episode of care if it
6 happened later and could be associated with that episode
7 of care?
8 A. No, unfortunately not, because if somebody had been in
9 hospital and then perhaps died at home a month later,
10 what you would do, you would notify the death to the
11 system. The only information, because it is an
12 administrative system, the information that goes on is
13 date of death, place of death, if known, and we would
14 normally make some sort of comment as to who had
15 notified us just in case there was a problem at a later
16 date. But it really is not linked back to any episode
17 of care as such.
18 Q. That would be, at that particular time for Patient
19 Administration Systems, the mechanisms that were used?
20 A. Yes.
21 Q. I was just wondering if there was any attempt made to
22 map them back. That would have been the standard
23 process at the time: the Patient Administration System
24 would not have linked the death to the episode?
25 MR LANGSTAFF: When you say "at the time", so we can have
0010
1 clarity here, we are looking at 1984 to 1995.
2 MISS HARDING: Yes. I think it is only in the last four or
3 five years that Patient Administration Systems, and
4 other systems, have been capable of making these more
5 sophisticated connections about being able to link
6 a death to a particular episode in a way that made it
7 clear that it was related to that episode, or that it
8 was not related to that episode but was related to some
9 other episode of care, or even not to any episode of
10 care that happened in the hospital. It is only recently
11 we have been able to do that. Would you agree?
12 A. I think so. I think now some of the systems are
13 becoming more sophisticated and you are getting into the
14 realms of patient's electronic records, then it has more
15 ability to link in the deaths with particular diagnostic
16 information, perhaps, in episodes of care. Because we
17 are talking about purely an administration system, it
18 had no links at all with diagnosis in any real
19 particular episode of care.
20 MISS HARDING: That is what I would have expected.
21 MR LANGSTAFF: One of the points you are making is that in
22 the system as it was operated in Bristol, the diagnostic
23 information was entered separately from the real-time
24 information which went on to the system?
25 A. Yes.
0011
1 Q. So how did the diagnostic information get on there?
2 A. The way that it works is obviously a patient comes into
3 hospital, they have their care, are discharged from the
4 hospital, and then the notes one would normally expect
5 to go back to the consultant's secretary to have
6 a summary dictated.
7 That summary should be dictated as quickly as
8 possible after the discharge, although some clinicians
9 are better than others at doing discharge summaries, so
10 the discharge summary would normally be a side of A4,
11 the top half would have the demographic information, the
12 middle portion would probably have the diagnostic
13 information written down, into a main and secondary
14 diagnosis. If they had an operation, there would be
15 a section for the operation details. At the bottom you
16 would have the text the medical staff dictated about the
17 patient's stay.
18 A copy of that discharge summary would obviously
19 go to the GP and any other interested parties. A copy
20 would be retained in the medical records. As soon as
21 that discharge summary has been dictated and typed,
22 those notes would then go off to the clinical coders for
23 that episode of care to be coded. That would be done as
24 quickly as the discharge summary was dictated. Most of
25 the coding clerks would code directly, all the coders
0012
1 would code from the diagnostic and operational
2 information that had been put on to that discharge
3 sheet.
4 MR LANGSTAFF: So the way it was done in Bristol was the
5 discharge summary was the basis of the diagnostic
6 coding?
7 A. Yes.
8 MR LANGSTAFF: Is this the way it is done generally or
9 not?
10 MISS HARDING: There are a number of different ways
11 it is done in different places. And some coders have
12 access to all of the case notes and do not have
13 a discharge summary and therefore would use all of the
14 information that was in the case notes rather than just
15 the discharge summary. In other places, coders just
16 have the discharge summary and do not have access to the
17 full set of case notes. In other places, coders code
18 from the discharge summary but being able to refer back
19 to the full set of case notes and any other information
20 they can use from the full case notes as part of their
21 process of making accurate diagnoses. So it works in
22 a number of different ways.
23 I would just like to ask a question, if I could,
24 about who did the discharge summaries?
25 A. Again, I could not say for sure, but it would be one of
0013
1 the consultant's medical team. I think normally one
2 would expect it probably to be the Registrar himself,
3 but that is not to say that the more junior staff did
4 not dictate summaries. I could not give you an accurate
5 answer on that.
6 What I was also going to say was that the
7 rationale for us giving the coders the notes was they
8 are very squirrel-like they take a big pride in their
9 work and, for instance, it is helpful if you have
10 a patient who contracts cancer to look back at the
11 histology report to input that data on to the cancer
12 screens. I am sure a lot of the coders have been around
13 a long while and will have the common sense to look at
14 appropriate sections of the notes to validate what they
15 are doing.
16 MR LANGSTAFF: Since you have raised it, can I explore
17 your perception of the quality of the coders here in
18 Bristol. First, really, we had some evidence yesterday
19 that there may be different types and qualities of
20 person interested in doing the work as a coder.
21 A. Yes.
22 Q. And we may have had an impression that coding perhaps
23 had not, at least in the 1980s and early 1990s, been
24 very highly valued, and for that reason, may not have
25 attracted the best quality of staff taken generally,
0014
1 although individuals may have been very good.
2 One indication of the quality of interest in the
3 job and the value people put on their own job may be the
4 length of time they actually stay in it and what sort of
5 staff turnaround you have. What was the picture at the
6 Bristol?
7 A. The frightening thing is, most of them have been around
8 as along as I have. As far as UBHT is concerned, most
9 of the coders that are in post now will have been in
10 post in the last 10, 15, 20, years. I cannot remember
11 the last time we had a coding vacancy within the Trust.
12 I think that is quite an important issue, because
13 they have not always been appreciated, in my view.
14 Their grades have been not abysmal, but they have not
15 been the best-paid staff within the Health Service.
16 I think they assumed a slightly higher profile perhaps
17 when contracts and GP funding came into the health
18 service where the information and diagnostic information
19 was used for charging purposes, and I think people then
20 started to sit up and say "The information the coders
21 are putting in is invaluable. We are basing our some of
22 costs on the coding", so they certainly assumed a higher
23 profile.
24 What we did as a Trust in the early 1990s,
25 I cannot be too precise about the date, the coding
0015
1 clerks were given to directorates in the Trust because
2 they were managing their own contracting. They were
3 centralised during the 1980s and very early 1990s but
4 they were given to the Directorate so they were made to
5 feel more part of the team they were working for.
6 Q. So in Bristol the person coding -- first of all, if I am
7 right in the thesis that I was advancing a moment or two
8 ago, suggesting that people would stay in a job if they
9 found some value in it -- there were people who, for
10 whatever reason, valued the work they were doing, and
11 secondly, so far as any specialty was concerned, they
12 were, as it were, specialist coders because you find
13 somebody in the cardiac surgery directorate, someone in
14 the children's services directorate, and so on?
15 A. Yes. From the time the staff were devolved to the
16 various directorates, it would have been the same two or
17 three coders coding the cardiac cases.
18 Q. So one would expect, if that is the case, a consistency
19 of coding approach?
20 A. Yes. Can I just say, prior to that they were in
21 a central office, and it was a sort of shared
22 responsibility, although what they tended to do when
23 they were in that central office, each one of them had
24 various responsibilities, so you may have one person who
25 was responsible for ophthalmology, somebody else was
0016
1 a bit of an expert in oncology, but at that time they
2 were managed, supervised, by a coding "expert" in
3 inverted commas, so she kept a very close eye on what
4 was happening in the section.
5 Q. It follows from that, as well, the next question I was
6 going to ask you, whether there was any moderation in
7 the sense of any checking to ensure consistency as
8 between coders and you did not get any great extent of
9 intercoder variation?
10 A. The audit of the coding has got better. Because the
11 coders were managed by the directorate, I have to say,
12 I do not think there was very much, if any, audit taking
13 place of what was being coded certainly in the early
14 1990s.
15 What we have done as a Trust in the last
16 18 months, we have actually appointed a Trust clinical
17 coding co-ordinator, because we wanted to ensure that
18 because the staff were diverse and being devolved to the
19 various directorates, some person is overseeing and
20 ensuring that all the national guidelines were being
21 implemented as and when required.
22 So audit coding I have to say probably was fairly
23 sparse in the 1980s and early 1990s.
24 Q. Is this the picture nationally?
25 MISS HARDING: I think to take a specific example of
0017
1 coders being associated with clinical directorates.
2 When clinical directorates were set up around the
3 management initiative, there was an incentive regionally
4 and nationally to make sure it was appropriate and more
5 timely. A number of different models were tried and
6 a model similar to the Bristol model, which was about
7 putting people where the specialists were, the theory
8 being they could talk more closely to the doctors about
9 the case notes, so bringing them away from the basement
10 to the garret, which is where traditionally they sat,
11 and actually making them more closely associated with
12 the directorates.
13 In some cases it worked very well. I think the
14 point you have just referred to, though, is that because
15 they were separated, sometimes the integrity and
16 reliability and the professionalism that you got from
17 having the coders working together actually was
18 sometimes lost.
19 I think the benefits of having them more
20 associated with the clinicians were actually the most
21 important thing, but supervision, co-ordination,
22 checking with each other, I think, was a bit lost by
23 that separation, but it was the model that was felt to
24 be most relevant at the time, and I think it will be the
25 model that we use about getting the clinical coders more
0018
1 closely involved with the case notes and the doctors who
2 were writing the information in the case notes. I think
3 that is fairly clear, that that was one of the things
4 that was beneficial.
5 If you have a large enough hospital as you have in
6 Bristol, there were two or three coders associated with
7 a particular directorate. Of course, that expertise
8 does build up with two or three very specific coders
9 becoming expert in those areas.
10 MR LANGSTAFF: Can I ask you to pause for a moment. We will
11 have your response first and then go back to
12 Ann Harding.
13 THE CHAIRMAN: Always remember that all of us, when we
14 get terribly interested in our topic, we sometimes tend
15 to speak slightly more quickly than normal. It is so
16 important that we do get a record. Can I urge you,
17 I realise there is an element of unnaturalness in it,
18 but if we could slow down just a bit, then we will not
19 lose anything.
20 MISS HARDING: I apologise.
21 A. I was going to say, when the staff were devolved to the
22 directorates, it depended to a degree on who their local
23 manager was as to what sort of supervision and help they
24 got with the coding. I think there were some cases
25 where they were very much left on their own and there
0019
1 were other people that were supervising the management
2 of those staff. They had an interest and knowledge of
3 the coding, and some of them through to successfully
4 coding, and I think you need to try and strike the right
5 note, really.
6 MR LANGSTAFF: You were going to go on?
7 MISS HARDING: I just wondered, given some of your coders
8 were working in very specialist areas, whether they were
9 able to talk to people, coding in those areas in other
10 hospitals.
11 A. I think that was probably one of our failures, that
12 they were given to the directorates. They were left
13 very much to those directorates to supervise and manage
14 them. It has only really been the last two, perhaps
15 three years, where we have tried to arrange regular
16 meetings with the coding staff so that it will give them
17 an opportunity to all get together.
18 Now we have a Trust clinical coding co-ordinator,
19 she has regular meetings with all the coders every
20 couple of months, to ensure that they all know who one
21 another are and that they are pulling and pushing in the
22 same way.
23 MR LANGSTAFF: Going back to the question of the quality
24 of the coders, what incentive did they have to get it
25 right?
0020
1 A. Their own personal satisfaction, I would suggest, no
2 more, no less. I think if you spoke to any of the
3 coders, not necessarily within Bristol but anywhere, and
4 Ann may very well support this statement, I have always
5 found coders to take a huge pride in what they do.
6 I have done it myself. I used to do coding quite
7 a number of years ago. I think if you look at the
8 discharge summary and you look at what the doctors put
9 on there, you can quite often get a feel as to whether
10 there is anything actually missing from that discharge
11 summary.
12 I can remember doing it myself. I used to ferret
13 through the notes and just see if I could pick up any
14 additional bits of information I felt to be appropriate
15 as perhaps a secondary diagnosis, like somebody with
16 diabetes. The medical staff quite often used to put on
17 the main diagnosis and forget about any other ailments
18 that the patient had which took them to hospital.
19 I think they take a pride in their work, they have been
20 around a long while and they are beginning to assume
21 a higher profile. I think people are beginning to
22 realise how important they are to the organisation.
23 MISS HARDING: I would support that. I think if
24 I could describe them, they are probably squirrels for
25 information. They do take enormous pride in trying to
0021
1 make sure that everything relevant does get put down.
2 I think the difficulty they have, although they
3 have become experts in their own area and require an
4 awful lot of clinical knowledge, I think the difficulty
5 they have is that the information that they have is all
6 that they can use. This is the point about them being
7 close enough to the clinical staff to be able to go and
8 ask them, "Is there something you have forgotten to
9 write down?" and "This does not seem to quite make
10 sense", but I think they are very professional in their
11 own right as clinical coders in the way that you have
12 described. I think it is amazing that there is a post
13 with no professional support, nothing that defines them
14 as a profession, that we have had such professional
15 support from in the particular group of experts.
16 I think the situation that is apparent from
17 Bristol would be mirrored across the entire country.
18 I do not think it would be substantially different
19 anywhere.
20 MR LANGSTAFF: So what you are saying is that despite
21 some of the reservations that were expressed yesterday,
22 you from your personal experience here, and you from
23 your experience and understanding of the national scene,
24 would actually think that coders on the whole do a good,
25 in terms of accurate and dedicated, job; they are not
0022
1 simply, as it were, filing clerks who wanted to get one
2 task out of the way?
3 A. No, I think they do not. They are dedicated staff.
4 MISS HARDING: I would agree. Obviously one cannot
5 generalise. The question you asked earlier on about
6 turnover of staff I think is very relevant here, that
7 staff do stay. The ones who do not enjoy it, do not see
8 it as valuable or find it very boring, go pretty
9 quickly. But we have this cohort of staff who have been
10 around for a long time and are very experienced.
11 THE CHAIRMAN: May I ask a question on that? Is it that
12 they stay because they are as you described them, or
13 because they have acquired a particular skill which is
14 not transferable to any other context?
15 MISS HARDING: I think it is true that the skill is not
16 transferable into another context. I think that is
17 absolutely true, but their general administrative skills
18 and general sort of I think organisational skills would
19 make those skills transferable into another area of
20 administration in a hospital, and given that they are
21 not very well paid, I think there would be an incentive
22 to move somewhere else for other pay, but that does not
23 happen terribly often.
24 MR LANGSTAFF: I suppose one answer to the point might
25 be, that would be a very valid point for someone
0023
1 starting off, who would think to themselves after 4, 5,
2 6, 8 weeks, "I do not want to do this for the rest of my
3 life and I will not be able to do anything else if
4 I stay, so I shall move on now". You are saying that
5 does happen, but not very often?
6 A. I think someone would know if they would be doing
7 coding for the next 10 or 15 years. It takes so long to
8 train somebody correctly and properly, it could take
9 anything up to 12, 18 months. You are always finding
10 new codes you are not quite sure what to do with. It is
11 important you have somebody you can ask what needs to be
12 done. I think because of the length of the training
13 period for coders, you would know very quickly if you
14 did not want to do that for the next few years.
15 MR LANGSTAFF: How does the system generally cope with
16 the learning curve of the coder?
17 A. What we have done now and what used to happen in the
18 past, if there is a new person appointed, they are now
19 linked to the clinical coding co-ordinator, so they will
20 spend as long as they need to be trained up to coding in
21 the areas they are responsible for, so she is like,
22 I suppose a nursemaid, if you like: she stays with them
23 or trains them thoroughly in the process of coding.
24 Some people pick them up quicker than others. I would
25 say it takes at least 6, 12 months to cope efficiently.
0024
1 MISS HARDING: I agree. Would it help if I talked about
2 the national picture around the support of clinical
3 coding and training?
4 THE CHAIRMAN: Mrs Maclean has a question.
5 MRS MACLEAN: Before we go on, I am very interested to
6 hear that with the new situation of the coders in the
7 directorates, they should have better access to query
8 matters they are unsure about with the clinicians.
9 I just wonder whether the queries came from the other
10 direction too: would a clinician ever query what a coder
11 was --
12 A. I think they would possibly query it if they were
13 interested. For instance, if perhaps a clinician came
14 to look at some information and wanted a particular
15 code, for instance, or there were patients who had
16 a particular procedure carried out, he would get that
17 information back from the systems via the coding. When
18 they start to analyse that situation, I expect there
19 would be occasions when he would find it had been coded
20 incorrectly or not quite as he would have anticipated.
21 I think by analysing and looking back at the coding,
22 that is when those questions would most likely crop up.
23 I have to say, I cannot remember too many
24 occasions where clinicians have actually sat down and
25 looked at what has been coded.
0025
1 MRS MACLEAN: Thank you.
2 MISS HARDING: Could I just add something to that?
3 I think this is one of the problems that we have, the
4 data is collected for the purposes which clinicians
5 believe is managerial, and therefore not relevant to
6 them, and I have a great deal of sympathy for that,
7 because the level of detail at which a clinician would
8 want the information for the purposes of audit is not
9 readily encompassed within the levels of diagnosis and
10 operative coding that we currently have.
11 I also think that we have been quite lax in not
12 feeding back to clinicians the information that is being
13 submitted to the central returns about what it is they
14 are doing.
15 When we did give information back to clinicians,
16 they said "I do not do that operation, I do not have
17 that many patients with that diagnosis", and I think
18 therefore one of the things we must do is to find ways
19 of ensuring that the information does go back to them so
20 they in some ways validate it before it is used for
21 other purposes.
22 The period we are talking about, I think that was
23 done for specific audits, and in some places was done
24 quite consistently, but only on a very small scale, if
25 at all. I do not believe that Bristol is any different
0026
1 from other places in that.
2 But your point, when people are given it back,
3 they do have very specific and detailed comments to make
4 on it, which are very helpful in improving both the
5 quality and thereby the usefulness of the information.
6 MR LANGSTAFF: So if we are talking about feedback, as
7 I think we are, then the feedback comes in at least one
8 way which is from the consultant or -- for the
9 consultant's own purposes, somewhat imperfect from what
10 you are saying?
11 MISS HARDING: Yes.
12 MR LANGSTAFF: After the purchaser/provider split and
13 when contracting came in, was there rather greater
14 feedback, at least for contracting purposes, than there
15 had been before?
16 A. I think certainly from an administrative perspective,
17 there would have been a lot more interest in the
18 diagnostic information, the coding that had taken
19 place. I am still not totally convinced at that time
20 there was much more interest from the medical staff, but
21 I think because a lot of the contracting was based on
22 the diagnostic information that was input on to the
23 system, certainly from perhaps a finance and
24 administrative point of view, a lot more interest was
25 taken in the information going in and out of the system.
0027
1 Q. In the diagnostic or in the operative? One appreciates
2 that operative procedures may cost very different sums
3 and therefore may have a contracting implication in
4 a way that diagnosis may not?
5 A. I think in both -- I guess it depends on the various
6 specialties, but I think probably -- I am totally
7 convinced there was a bigger interest in the diagnostic
8 and the operative information when contracting came in.
9 MISS HARDING: I would agree. I think I also agree with
10 you that really did not go back into the clinical
11 arena. I think the difficulty we have is that the data
12 we collected was for administrative purposes and that
13 the new focus on clinical information for clinical
14 purposes, clinical governance, is the thing that will
15 drive the information going back to the clinician, but
16 I think that although that has happened, probably in
17 a small number of cases, I still do not think that is
18 the norm for data to be fed back to clinicians, even
19 resulting from the contracting process.
20 MR LANGSTAFF: But nonetheless, did the feedback from
21 the contracting process have an impact on the care or
22 the quality of the coding, do you think?
23 A. I think probably not.
24 Q. Because it was so good in the first place?
25 A. Absolutely.
0028
1 MISS HARDING: I think in some places where the standard
2 of coding was already quite good, there probably was not
3 such an incentive to improve it, but I think in other
4 places where the quality of the coding was not very
5 good, particularly in terms of operative procedures,
6 then I think there was much more attention paid to the
7 quality of the coding associated with contracting than
8 it ever had been before.
9 MR LANGSTAFF: One would otherwise find it very difficult
10 to explain what we were looking at yesterday, for
11 instance, when we came to looking at HES and the way it
12 appeared that coverage and particularly completeness of
13 data improved almost dramatically after 1991/92.
14 A. That would have been again for contracting purposes,
15 because obviously the information had to be put in
16 within a certain period of time, otherwise you were
17 potentially losing money. So I think you will find that
18 is why there was probably a more timely inputting of the
19 data. I think that the local managers would have moaned
20 to the clinicians if they were not getting the discharge
21 summary out in a timely fashion to allow the coding to
22 take place.
23 MR LANGSTAFF: So to some extent, we may actually be able
24 to achieve more from looking at PAS locally than one
25 could tell from taking the HES statistics in respect of
0029
1 Bristol, because, at some stage, PAS being dynamic in
2 the information, the diagnostic coding would have been
3 put on to it and it may not have been reflected in HES
4 because the discharge summary had not been produced, the
5 clinician had not yet tidied up his case notes a couple
6 of months before the discharge summary goes out,
7 whatever it may be.
8 MISS HARDING: I think it is quite difficult to quantify
9 how different PAS would be from HES. I mean, it should
10 not be different, but we know in different places that
11 we were able to submit further data to HES to update the
12 original submissions, but that did not always happen for
13 quite complicated reasons, that it is actually quite
14 difficult to update the structure of the database in
15 a way that allows it easily, and then extract that data
16 and load it back up to the HES data.
17 So I think that does differ from place to place
18 about how different the PAS data is to the HES data, and
19 I would be interested to know what your views about how
20 that would be in Bristol?
21 A. Again, it is difficult for me to say, I think. I think
22 you would need to speak to one of our information people
23 to clarify the situation with regard to that. It is
24 best if I do not comment, I do not think.
25 MR LANGSTAFF: We have looked, then, at the coders and the
0030
1 general satisfaction expressed as to coding and the
2 quality of coding particularly, I think, in the period
3 from 1990 onwards. We can draw from PAS probably quite
4 a lot of information about deaths, although I am not
5 sure we have explored how accurate that would
6 necessarily be. Obviously it depends on input which
7 comes in, you were saying, not only the hospital input,
8 if there is a death in hospital that is easy, but if
9 there is a death after, connected with the operation,
10 I think we have established that there is no link?
11 A. That is right. The problem we have, say someone came
12 from Cornwall, came to hospital, was discharged back
13 home and died six months later.
14 If the patient does not have to come back to the
15 hospital, there is not a lot of incentive for the
16 relative or GP's surgery to tell us that patient has
17 died. We are very dependent on outside agencies letting
18 us know if patients have deceased. It is only as
19 accurate as the last time the patient attended, to
20 a degree, with the caveat that the GP's surgery or
21 somebody else may have given us information.
22 MR LANGSTAFF: May I take this back in stages from where we
23 are now? Nowadays, with electronic systems, do we find
24 out electronically when people have died, so that the
25 database can be updated?
0031
1 MISS HARDING: Not yet.
2 MR LANGSTAFF: So that is coming?
3 MISS HARDING: That will come with the next stage of the
4 information strategy.
5 A. The links to the local health authorities and other
6 agencies that deal with health, those links are in the
7 foreseeable future. We are not any way from having
8 linked with the local Health Authority, we have been
9 having links with them in the last few months trying to
10 find a more seamless way of trying to make our data
11 similar, if not the same.
12 MR LANGSTAFF: So if someone were to die five years
13 after surgery, leaving aside whether there is any
14 connection or not, that may not be picked up unless
15 there is a continuing hospital attendance?
16 A. There will be a lot of patients on a Patient
17 Administration System who have had inpatient care that
18 will have died after they have been discharged and the
19 system will not know about it; there will a lot of those
20 patients.
21 Q. What about the patient in respect of whom there is some
22 follow-up?
23 MISS HARDING: I think there are some follow-ups that have
24 been done for audit processes. I think that has been
25 very useful with consultants, and again, the importance
0032
1 about the audit process is that it is clinical-led, so
2 there would be some carry-through from the audit process
3 with individual consultants wanting to follow patients
4 through, sometimes in conjunction with the general
5 practitioner as well.
6 There is one other source of data that can be used
7 to look at deaths, and those are the death registrations
8 and the analysis of that that is available through the
9 Office of National Statistics and a source of data
10 called the "death takes". That may be one of the things
11 that could be considered as following through individual
12 patients. But that is not something that is done
13 routinely.
14 A. No, and of course at the minute that information could
15 not be reconciled to the inpatient stays; our system
16 still cannot currently do that.
17 MR LANGSTAFF: For the patient who was still subject to
18 review in hospital at six-monthly or yearly intervals,
19 let us suppose, what systems were in place in the 1980s
20 and 1990s to prevent the hospital sending out an
21 outpatient attendance note, an appointment, to somebody
22 who had died and thereby perhaps unwittingly causing
23 offence to relatives?
24 A. It happens. I mean, I can give you a number of
25 occasions where patients have not turned up for an
0033
1 outpatient appointment and we have written to them
2 a polite letter saying "You had an appointment, you did
3 not attend, would you like to come on another date?"
4 offer them a date, and a distraught relative will say
5 "It is unfortunate my mother/son/daughter passed away
6 several weeks/months ago". That relates back to the
7 point I was making just now: the information is only as
8 good as the last time the patient attended or the last
9 time the GP or some other agency told us that a patient
10 was deceased.
11 It does happen. We have no seamless way of
12 updating the database with regards to death information.
13 Q. How good would you be at disclosing information?
14 MISS HARDING: At this particular stage, not terribly
15 good. There was no automatic way of doing it. Since
16 then we have had the development of an automatic way of
17 updating information into the master patient indexes in
18 hospitals and other places from the national
19 administrative database so that has improved things, but
20 at this period of time, it was not accurate and it was
21 not timely, and not complete.
22 A. One advantage we did have, which perhaps may not have
23 been appropriate at another Trust: because it was
24 a Trust system, if a patient has come into the BRI for
25 some sort of surgery but has also been attending the Eye
0034
1 Hospital or the Dental Hospital, if they had been
2 informed the patient had died, it would update the Trust
3 system. So that is one big advantage of having a Trust
4 system: it updates the system for the whole of the
5 Trust, it does not matter which hospital it is. So
6 I guess in a way, the fact we have more hospitals on the
7 system, we are slightly increasing our chances of
8 getting the information.
9 MR LANGSTAFF: So if one is to compare the PAS system
10 with one coming from HES or clinical records, one would
11 expect that PAS would have better information about
12 death, but it is still very patchy?
13 A. Yes, I think that is a good summarisation, yes.
14 Q. So looking at what use this Inquiry can make of the
15 information which is supplied by PAS in the way that you
16 have told us it was collected, one would expect the PAS
17 system to reflect the information available in the
18 clinical records, would you not?
19 A. Yes. I mean, as I say, the PAS is real-time. We have
20 a number of in-built checks to ensure that patients are
21 put on to the system and the appropriate transfer and
22 discharge information goes on as well. Therefore,
23 I would expect every patient who has come into the
24 hospital to be on the PAS system.
25 Q. So if clinical notes, for instance, for whatever reason,
0035
1 went missing, there would still be a record on the PAS
2 system?
3 A. Yes. I mean, the PAS system, things like registration
4 numbers, the patient's hospital registration number is
5 held on the PAS system. If you can find the notes, you
6 can search the system by name, date of birth. You can
7 at least find the patient you are looking for, find
8 their hospital number, although you may not be able to
9 find the notes. You could then dip into the summary of
10 the episodal information and get things like dates of
11 admission, dates of discharge, any other information you
12 wanted relating to that episode.
13 Q. So there is a cross-check available between the clinical
14 records and the PAS, and to the extent that the
15 cross-check shows that they reveal the same things, one
16 would have greater confidence in information derived
17 from either source?
18 MISS HARDING: Yes. I think that is right. I think that
19 one needs to be quite careful about chronology, because
20 the patient record itself would have a series of events
21 in it that would be within one episode. Depending on
22 the importance of those, some of those would then be
23 coded through to a diagnostic code and a operative code,
24 but I think with complicated interventions and long
25 episodes of care, then one would need to think through
0036
1 very carefully how much of that information would be
2 transferred not from PAS but from PAS to HES, I think.
3 That is one of the things one would need to look at.
4 A. We have already extracted from the PAS system a lot of
5 episodal type information for the Inquiry to look at
6 that has been requested over the period of the last few
7 months.
8 MR LANGSTAFF: Obviously there is a check as to the
9 completeness of the case records we have by
10 cross-checking with the PAS system. It also gives
11 information as to what the coders, the experienced
12 coders that you have described, classified the various
13 operations as being and one can cross-check that against
14 any in-house exercise to code the same records by other
15 coders. So there is a degree of cross-validation
16 available?
17 A. Yes.
18 MR LANGSTAFF: I do not know, Dr Spiegelhalter, whether
19 you want to comment at this stage? Whether there is any
20 question you wish to ask from your perspective?
21 DR SPIEGELHALTER: It builds on where the discussion had
22 got to, because PAS forms the intermediary link between
23 what is in the clinical records and what eventually gets
24 on to HES and in this Inquiry we are going to be
25 recoding the clinical records and looking at the HES
0037
1 data, so the strength of that link, the accuracy of
2 those links, would be looked at in detail for Bristol
3 and hopefully, from what you are saying, reasonably good
4 correspondence will be there, which will be very
5 encouraging.
6 For the other centres that we would like to
7 compare Bristol with, all we are going to have is the
8 HES data, and I suppose -- this is perhaps outside what
9 you would be willing to discuss, but may be Ann could
10 say something. If a good correspondence was found
11 between the records which one could think was the gold
12 standard in HES in Bristol, do you think that degree of
13 correspondence is likely to have occurred after 1991 in
14 other centres in the country?
15 A. I sit on various regional and national groups. I can
16 certainly speak for the South West region, which is
17 Cheltenham down to Cornwall, excluding Wiltshire as it
18 was then, because that was not part of the real South
19 West. All of those Trusts as they are now within the
20 South West region all have the same software. They are
21 all using the regional computer centre's PAS system.
22 I used to meet regularly with my colleagues from
23 around the region and I am reasonably confident that
24 most people, most hospitals, were using the system in
25 very similar way.
0038
1 The functionality was obviously identical. It was
2 the same system, so the functionality was absolutely
3 identical. At that time, there was no local deviation
4 from the regional system. You had the system and it was
5 agreed, regionally we used to meet regularly as a group
6 to enhance and change the system. We used to change the
7 system as a group, so I am totally confident everybody
8 was using the same system.
9 What I could not be totally sure of is where and
10 how people were inputting data to the system. I am
11 totally confident that most hospitals in the region were
12 inputting their admissions and discharges in real-time.
13 I could not make any comment about how they were doing
14 the coding, but most people were using it in a very
15 useful fashion in the South West. I cannot make comment
16 about outside the South West, because although I have
17 seen other systems and visited other hospitals, I have
18 never spent enough time there to comment on how they
19 were or were not using the systems.
20 MISS HARDING: I think Bristol at this time, and in the
21 South West, probably had a very good Patient
22 Administration System. The fact that everything was
23 being entered in real-time was certainly not standard in
24 other places. Although other people had Patient
25 Administration Systems which you could use in
0039
1 real-time --
2 MR LANGSTAFF: Could I ask you to pause there and take the
3 microphone closer to you?
4 MISS HARDING: Although it was possible to use other
5 Patient Administration Systems in real-time, a number of
6 places actually tried it and decided that it did not
7 work effectively enough in the particular way they set
8 up their hospitals. One of the strengths Bristol had
9 was that it was used in real-time.
10 The other point is that there were not that many
11 Patient Administration Systems around. Their history
12 was coming from regional computer centres and being
13 owned by regional computer centres in the 1980s before
14 they were made commercial as part of the newer way of
15 working.
16 It would be possible to look at the way other
17 Patient Administration Systems were working to make
18 comparisons about (a) whether they were live systems;
19 and (b) whether that made any difference to the quality
20 of data that was put through from PAS into the HES
21 system.
22 That could be done because there were some
23 standard monitoring routines that we used to compare the
24 HES data and the PAS data in the 1980s and 1990s, and we
25 had something called a "metadata file" that it might be
0040
1 possible to look at. Again, I do not know, but there
2 were processes put in place to try and check between the
3 two sources, PAS data and HES data, sometimes on
4 a regional basis, depending where the PAS system was
5 owned, but at other times at health authority level as
6 well. It may be possible to look at that.
7 A. I think it was unusual to have a system that the users
8 were able to influence how it changed, along with local
9 and national guidelines. I think that was one of the
10 huge benefits of the system; it really was enhanced and
11 taken forward by the users. It was the way it was set
12 up within the region. I think also the way we
13 implemented the systems eventually within the UBHT,
14 I think our view was that we needed to implement the
15 systems gradually because it was a huge culture shock
16 and a change for most of the staff, and I think, I am
17 totally convinced when we implemented various aspects of
18 the system, we made sure that we could take the staff
19 along with us. I think that has been an important
20 ingredient in making sure that the quality of the data
21 is good. We could have done it in 6 months and ended up
22 with rubbish going in and rubbish coming out. We did it
23 slowly and gradually and took the users along with us.
24 MISS HARDING: I think that is a strength. I think the
25 reason other people who were trying to implement
0041
1 real-time Patient Administration Systems actually gave
2 up is because the culture and their organisational
3 issues were so great that they were not able to do that
4 at that particular time.
5 A. I think, you know, you kind of build up, you have to
6 take the users along with you. That is one of the
7 things we aimed to do right from the very beginning:
8 build up a core of expertise and enthusiasm, and then
9 you can take the staff along with you. I like to think
10 we have been fairly successful in achieving that.
11 MISS HARDING: My only comment would be that that was
12 for the administrative data rather than the clinical
13 data. We go back to the point that the purpose of the
14 PAS is an administrative system and not a clinical
15 system.
16 A. Yes. That is why the clinicians have not always seen
17 it as anything that can help them. They have seen it as
18 the administration of the hospital and the Trust and it
19 does not give them very much. I think that has been the
20 view of the system.
21 Q. And that would be mirrored elsewhere?
22 A. Yes.
23 DR SPIEGELHALTER: Going back to something Mr Willmer
24 said yesterday, recently the government has released
25 performance indicators derived from the HES data, which
0042
1 I wonder whether those will focus clinicians' attention
2 much more on the administrative systems, because the
3 variability between institutions was shown to be very
4 large, and probably a lot of that was due to faulty
5 information systems.
6 A. Hopefully it will help them to focus their minds on
7 some of these issues.
8 MISS HARDING: I think this is a very, very important
9 point, and some of the subsequent analysis to the
10 original analysis of the performance indicators that was
11 done in order to refine the accuracy of the original
12 tables of indicators has been very helpful. A number of
13 clinicians have been involved in looking at the data to
14 try and identify those issues where the data was just
15 wrong, and wrong in a sort of comprehensive way rather
16 than just incorrect, but wrong and wrongly used.
17 I think that has helped enormously in heightening the
18 understanding of (a) the performance indicators; but
19 (b) the need for good quality information that is owned
20 by clinicians rather than owned by an administrative
21 process. That has been very important already.
22 MR LANGSTAFF: In the course of analysis of the data
23 locally, would one expect, Dr Spiegelhalter, to throw up
24 any inaccuracies in the data, in the same way as was
25 described yesterday in respect of the HES system?
0043
1 DR SPIEGELHALTER: I think that the exercise that is going
2 on at the moment was the recoding of 1800 records and
3 the ability to compare those with the HES system and
4 temporary HES data will be an unprecedented opportunity
5 to see exactly how the system was worked in the past and
6 to judge the reliability of the way in which the HES
7 data has been derived through those stages of coding and
8 their derivation from the PAS system.
9 MR LANGSTAFF: Having heard what you have heard about the
10 way in which things were done in Bristol and got some
11 idea of how things might be done elsewhere, would the
12 fact of a general correspondence between the results of
13 the recoding of those clinical records in respect of
14 Bristol being similar to the results one would get from
15 looking at PAS and HES in respect of Bristol, give one
16 any confidence or additional confidence that one might
17 expect the same elsewhere when one comes to make
18 comparisons?
19 DR SPIEGELHALTER: That is really why I was asking whether,
20 if such correspondence occurs, is it because Bristol was
21 the only place in the country that was doing it
22 properly, or whether that is an extendable lesson and
23 something where we would have to ask some further
24 opinion?
25 I think even with Bristol, even if essentially the
0044
1 denominators are correct in terms of what cases were
2 being processed as accurately reflecting HES, I think
3 the problem with the HES that it will only contain
4 really the in-hospital mortality, will be a problem that
5 is unavoidable and will inevitably involve that HES will
6 miss some of the deaths, but we will know for Bristol
7 exactly the extent to which that is the case.
8 MR LANGSTAFF: There may be two respects in what you have
9 been saying in which the PAS system will provide better
10 quality information potentially than the recoding
11 exercise, and I put these out for comment: the first of
12 them is the fact that the coders were few in number,
13 specialists in their area and were able to talk in the
14 directorate structure to the clinicians means that they
15 had access to the clinicians in a way in which those
16 doing the recoding exercise have not had, and that
17 potentially may make a difference?
18 A. Yes. I think perhaps it has to be said that, yes, they
19 were within the directorates. I am not too sure how
20 often the clinicians would have gone to the coders to
21 make comments about how they were going to code.
22 I suspect that the coders would only talk to a clinician
23 if there was something they thought was missing or
24 perhaps was fairly bizarre. I would not want you to get
25 the impression that the clinical staff were popping into
0045
1 the coding room every day or every other day. That
2 certainly is not the case. I think that needs to be
3 made quite clear.
4 MR LANGSTAFF: You have answered that one. The next is that
5 it is possible for a PAS system to show us rather more
6 records than we may have from the clinical records,
7 despite the best efforts of the Trust to give us all the
8 records.
9 The reason I put that as a possibility arises from
10 a letter I am going to show you on the screen now. It
11 is UBHT 154/190. It is a letter from you on 14th June
12 1993 to Mr Wisheart.
13 We can see your name at the bottom. If we just
14 read through it, you are explaining to Mr Wisheart that
15 the notes of patients who had open heart surgery are not
16 being destroyed as part of the BRI's destruction policy
17 for medical records. Then in the middle paragraph, you
18 mention that thousands of sets of notes which had been
19 filed in an old ward at Barrow Hospital got wet and
20 mouldy.
21 A. They became a health hazard, yes.
22 Q. And they had to be destroyed?
23 A. Yes.
24 Q. Do you know whether there were any records which might
25 relate to the period 1984 to 1995?
0046
1 A. I do not think there were. Obviously I have been in
2 constant discussion with the Inquiry team with regards
3 to the medical records that they required, and I think
4 it would be safe and fair to say that we have managed to
5 find a huge proportion of the notes that have been
6 filed. I think there are probably only a handful still
7 missing.
8 It was interesting when Mr Wisheart contacted me
9 about the destruction policy we had, it was a very, very
10 comprehensive policy and things like open heart surgery
11 were excluded from it. Again, I could not say with my
12 hand on my heart that among the notes at Barrow, there
13 were not one or two that related to the Inquiry. They
14 were significantly older records than the ones I think
15 you are interested in, and they were located in an old
16 ward out there where the roof had started to leak and
17 they were in a disgusting state.
18 Q. So essentially you are confident that perhaps bar one or
19 two, but only a minimal number, we have all the records?
20 A. I am confident that there were hardly any notes, if any
21 at all, in that old store you would be interested in.
22 In fact, a lot of them, we could read the names and the
23 numbers on, we have actually documented the names and
24 the numbers in case there have been any enquiries on
25 those, so most of them we managed to get a bit of paper
0047
1 with the patient's basic details on so we could document
2 what we were actually throwing away.
3 Q. So it follows, I think, from what I have been asking you
4 about, that one could use the PAS system to check on the
5 completeness of the clinical records that have been
6 provided to us?
7 A. Yes.
8 Q. And equally, to check on the completeness of the
9 surgeon's own personal logs of the operations they did,
10 and any other logs which there may be in-house?
11 A. Yes, because when we were pulling the medical records
12 for the Inquiry, our first target was the doctors'
13 logs. We went through Mr Dhasmana's and Mr Wisheart's
14 logbooks to get the patients' information, and they went
15 from 1984 to 1995, if my memory serves me correctly.
16 Our first pull of the medical records was from those
17 logbooks.
18 Subsequent to that, one of our information
19 officers then went through the Patient Administration
20 System to look for the patients who fell within the
21 deadlines and the right specialty. We did actually find
22 some that were not in the logbook. That has been
23 documented in a note I sent through John Gray back to
24 the Inquiry.
25 Q. Yes, and thank you for that. Is there anything else you
0048
1 would wish to ask, Dr Spiegelhalter, as to the utility
2 that the Inquiry can make of the PAS system as it has
3 been described today?
4 A. No.
5 Q. In which case, can I thank you very much for what you
6 have said thus far and for Miss Harding's contribution.
7 Is there anything you would like to add? Every witness
8 before has been given that opportunity and there is
9 absolutely no reason why you should not have it as
10 well.
11 A. No, I do not think so. Most of the things I wanted to
12 say I said in my statement originally, so there is
13 nothing I wish to add.
14 MR LANGSTAFF: There may be some questions from the Panel.
15 THE CHAIRMAN: Yes, there are. First, Mrs Maclean.
16 Examined by THE PANEL:
17 MRS MACLEAN: I am looking at your written statement, the
18 last page, paragraph 21, WIT 211/7. We may not need to
19 call it up. I can read it to you. It simply says:
20 "It is important to stress that until the early
21 1990s, very few people in the Trust were familiar with
22 computers and their uses in audit or maintenance of
23 administrative data and sharing information."
24 I wonder if you could elaborate a little on that,
25 perhaps in the context of other hospitals in the South
0049
1 West which you were familiar with?
2 A. As has already been stated, I came back to the Trust in
3 mid-1988 and my primary responsibility at that stage was
4 to implement the PAS system right across the Health
5 Authority as it was then.
6 I had a remit which spanned a period of three or
7 four years and that remit was to implement four aspects
8 of the system: the index, inpatient module, outpatients
9 and waiting lists, and I had to have a plan of campaign,
10 if you like, to make sure that I could achieve that
11 target.
12 But I soon realised that because I was coming back
13 to a Health Authority where most of the staff that would
14 be using the system initially had no real experience of
15 using a keyboard or a computer, so as I again said
16 a short while ago, we tried a "softly, softly, slowly,
17 slowly" approach to make sure we got everyone on board.
18 The implementation was gradual and the training of the
19 staff was gradual, so when, for instance, we set the
20 index up, there were only a small number of staff that
21 needed the facility to register and change patient
22 details. We then moved on from that to the inpatient
23 module and that brought on board another bundle of
24 staff, ward staff, admission staff, staff who were
25 responsible for the non-admission transfer discharge
0050
1 detail and we spent as along as we needed with them to
2 make sure (a) they could use the terminal, (b) they
3 understood what we were trying to do and (c) they
4 actually used the system efficiently in the manner we
5 wanted them to do.
6 The same applied to outpatients and waiting
7 lists. We gradually took everybody on board. One or
8 two people tried to run away from me, I caught up with
9 them all in the end, but the "proof of the pudding" is
10 in the staff. If you spoke to most of the staff in the
11 Trust, they would not go back to the way we used to
12 work. It is much more efficient than the manual systems
13 and certainly helps when we are trying to answer
14 patient-based enquiries.
15 MRS HOWARD: You made it clear that this was very much
16 an administrative system, and we have also heard about
17 what I would phrase the "lack of ownership" from
18 clinical staff with regards to that. Do you have any
19 comment about subsequent setting up of stand-alone
20 systems within the Trust to support the clinical staff
21 in the information that they wished to collect? That
22 would be the first part of the question. I would like
23 to explore that after your answer.
24 A. Certainly, as far as I am concerned, I have only ever
25 been involved in implementing the corporate systems.
0051
1 Again, I think probably the Trust philosophy has always
2 been, where possible, we would like to use those
3 corporate systems, but that is not to say that if
4 a clinician or a group of clinicians wanted to go and
5 purchase perhaps an audit system -- they should be able
6 to do that.
7 I think the problem with doing that is, if you are
8 inputting the data into two separate systems, it is
9 always going to be difficult reconciling the information
10 on the two systems. We are trying to move as a Trust
11 towards integrated systems. We have the ability already
12 for the PATS system, for instance, to extract some
13 information from our corporate systems. That sort of
14 information is currently in a minority within the Trust.
15 Q. That anticipates my second question, which was the idea
16 of integration. Were there any examples of that during
17 the period of our Inquiry, of attempts to integrate the
18 PAS data with clinical systems?
19 A. Could you repeat the middle bit of the question?
20 Q. During the period of the Inquiry, were there any
21 attempts or examples of attempts by clinical staff to
22 bring together the PAS system and stand-alone systems?
23 I am particularly interested obviously in the systems we
24 have heard about, such as METASA and PAS.
25 A. I cannot really comment, because if clinicians or other
0052
1 staff within the Trust were interested in the
2 information side of the system they would talk to our
3 information people rather than myself. If somebody had
4 asked me for some information, I would point them in the
5 direction of the appropriate person. That was not
6 really part of my remit.
7 Q. So information and medical records were separate and
8 distinct?
9 A. No, not medical records. I actually manage the Medical
10 Records Department at the Bristol Royal Infirmary. I am
11 also the adviser for the Trust. The information section
12 lives within the IM&T department, but it is a separate
13 section.
14 MRS HOWARD: Thank you.
15 THE CHAIRMAN: There are no further questions from the
16 Panel. Mr Chambers?
17 MR CHAMBERS: I have no questions.
18 THE CHAIRMAN: I am grateful.
19 Mr Langstaff, you have already said, I think (or
20 if you did not, I will say) to Mr Hooper, in addition to
21 our thanks, that if there are any other matters that you
22 wish to bring to our attention as time goes on, we will
23 always of course be anxious to hear from you. You have
24 been very helpful this morning. We are very grateful to
25 you, and equally to our two experts, Miss Harding and
0053
1 Dr Spiegelhalter, from whom we will hear after a break.
2 Thank you, both of you, also. What I suggest,
3 Mr Langstaff, is that we take a short break now, for let
4 us say 15 minutes, and reconvene, at which point I think
5 we are going to hear from Dr Spiegelhalter again?
6 MR LANGSTAFF: Yes. We will hear briefly first from
7 Ann Harding who wants to pick up a couple of matters
8 which were left over, if you remember, from yesterday by
9 Mr Willmer, and she is able to deal with those matters,
10 technically I think in evidence, so she will give
11 evidence briefly wearing, as it were, her "evidential"
12 hat rather than her "expert" hat.
13 THE CHAIRMAN: Forgive me, I anticipated you quite
14 wrongly. We will reconvene then in 15 minutes, thank
15 you.
16 (11.23 am)
17 (A short break)
18 (11.55 am)
19 MR LANGSTAFF: Sir, I had advertised in the brief commercial
20 for the second session this morning just before the
21 coffee break, two events. It is a correction of what
22 I said yesterday morning. If one looks back to the
23 transcript at page 10, line 21, I was talking about the
24 Trust PAS system. What I said was that the data was
25 generated by clinical coders using standard
0054
1 classification systems. I am told through the sources
2 available to us that when this was released on the
3 Internet and examined with the care with which our every
4 word obviously is, it was appreciated by a number who
5 read this that this was inaccurate because the word
6 "clinical" was not inserted before the word "data".
7 That of course is what I had in mind, and that is what
8 I should have said, and I failed in giving the
9 exposition at the beginning to indicate sufficiently
10 that the data I was talking about was of course clinical
11 data, because the data which goes onto the PAS system is
12 necessarily put on administratively, as we have heard in
13 the course of this morning.
14 I do not want there to be any confusion or
15 consternation about that. This shows the very valuable
16 way in which those who read our transcripts on the
17 Internet are actually able to feed back to us and to
18 identify, if there be errors, any errors. So it
19 actually demonstrates the validity of the public process
20 upon which we are engaged.
21 THE CHAIRMAN: Absolutely, Mr Langstaff, and I would
22 emphasise and re-emphasise that point.
23 MR LANGSTAFF: Miss Harding, would you stand to take
24 the oath, please?
25 MISS ANN HARDING (AFFIRMED):
0055
1 Examined by MR LANGSTAFF:
2 Q. We already know, Miss Harding, that you are Ann Harding,
3 the Acting Director of the NHS Information Authority?
4 A. Correct.
5 Q. You were here yesterday when Mr Willmer was giving his
6 evidence to us. There were a number of matters which he
7 felt he could not answer but others might. You are in
8 a position, I think, to give the answers?
9 A. Yes, I am.
10 Q. Would you like to do so?
11 A. Thank you very much indeed, I would. If I take the
12 questions as they were raised with Mr Willmer yesterday
13 and then provide an answer to them, some of the topics
14 that we discussed this morning have already covered some
15 of the questions that were asked yesterday, so I will
16 try not to cover that ground again. I think there were
17 one or two things maybe to add to what was said this
18 morning.
19 The first question was, is there an incentive,
20 motivation, to get coding right? We had some discussion
21 on that this morning. I think the biggest commitment is
22 the personal commitment of the coders, who see their
23 contribution to the provision of good quality
24 information as absolutely paramount in the work they
25 do. I think that is number 1 and very important.
0056
1 The second motivation I think is more recent than
2 in the 1990s, and that is through the resource
3 management initiative which is where the NHS started to
4 look at the costs and also subsequently the
5 effectiveness of treatments that were given and
6 therefore were more committed to understanding the need
7 and having good quality clinical information.
8 I think the use of the data that we have made in
9 the 1990s has made it much more important and therefore
10 the motivation for individual coders, but more
11 importantly I think their managers, to support good
12 quality coding has become much more relevant.
13 That was the answer to the first question.
14 The second point --
15 Q. Would you wait for one moment there, because again,
16 you are naturally talking fairly quickly.
17 A. I am sorry. The second question was: what training,
18 pay, status, do these people have? I think the training
19 aspect is one that I need to put in perspective on
20 a national basis, that there has always been, originally
21 in Manchester but then more recently in Loughborough, an
22 organisation which is responsible for the quality and
23 training of clinical coders related to the whole process
24 of the development of coding and classification schemes
25 in conjunction with the World Health Organisation, for
0057
1 the ICD classification, and with the Office of
2 Population Censuses, now the ONS, in terms of procedure
3 classifications.
4 The responsibility of the trainers has been to
5 ensure that good quality clinical coding is available to
6 the NHS to support its managerial and epidemiological
7 processes through the HES data and those staff that have
8 been through the training programmes have had the
9 benefit of nationally trained and accredited trainers as
10 their mentors during their own particular training.
11 I think that is an important aspect of the
12 professionalism that is seen by the clinical coders and
13 also by those responsible for analysing and providing
14 information based on clinical coding.
15 So that is the issue about training: I think
16 a much more structured process than people probably
17 recognise.
18 In terms of pay, there was a discussion this
19 morning. My view is that clinical coders are paid on an
20 administrative and clerical grade of the Whitby Council,
21 and an inexperienced coder would now be at grade 3
22 level, probably between œ12,000 and œ15,000.
23 Experienced coders would probably be moving up to
24 grade 4s and managers now at grade 6. I am afraid I do
25 not have exactly the pay scales that relate to those
0058
1 grades, but I think you can see that from the level of
2 responsibility they have, those people are not very well
3 paid.
4 If you go back to what they would have been paid
5 at the beginning of the Inquiry period, I think they
6 would have been paid very, very substantially small
7 salaries in relation to other people.
8 I think valued, but not rewarded in terms of
9 remuneration.
10 They would have been comparable with medical
11 records staff and not seen as having expertise above
12 records clerks, and the skills needed as we have
13 discussed are substantially different.
14 As to status, I think again this has changed in
15 the 1990s, but I think the concept of a clinical coder
16 sitting remotely in either basement or in the garret is
17 one that actually describes very realistically how they
18 were placed. So I think not a high status.
19 The next question was, can differences occur in
20 coding the same procedures in different ways? We always
21 have issues about reliability and validity between
22 coders coding something once and then coding it a second
23 time subsequently in another set of case notes, but
24 there are very explicit rules and conventions set out
25 both within the International Classification of Diseases
0059
1 documents and the OPCS documents as well about
2 convention and rules for coding, and coders are trained
3 in using those and also trained to ask when they do not
4 understand what those rules mean. I think primary and
5 secondary diagnoses, we do have ways of looking at
6 primary and secondary diagnoses and trying to ensure
7 that there is consistency about how they are used and
8 some of the other coding conventions.
9 But of course, and I will come back to this
10 a little later, they are entirely dependent in their
11 ability to code on the evidence available to them from
12 the case notes and coding sheets. I would like to
13 address that in a bit more detail in a moment.
14 Q. Can I pause there? Are you perhaps saying that the
15 variability which there may be between coders which
16 implicitly you may be acknowledging is not their fault;
17 it is somebody else's fault?
18 A. I think that there are sometimes differences that are
19 their fault. That is probably about inexperience and
20 perhaps inadequate training at times, but the bigger
21 variability is not their fault, it is on the information
22 they get to code.
23 Q. But in terms of looking at the finished product, which
24 we have to do, you are, I think, accepting a degree of
25 variability from whichever source?
0060
1 A. Yes. The next question that was asked is: what feedback
2 do coders get as to the accuracy of their work? Again,
3 I think from the beginning of the Inquiry period to the
4 end of the Inquiry period, this has changed
5 substantially, for some of the reasons I have already
6 outlined.
7 In terms of some very practical things that one
8 could do, and still can do, the submissions from the
9 Patient Administration Systems that are used to feed the
10 HES data, there were high level reports back about
11 accuracy, the number of outliers, which, in places where
12 I have been, have always been used to feed back
13 information to coders about the accuracy of the
14 submissions that have gone through to HES.
15 Q. Can I again just pause you to clarify what is meant?
16 You have used the expression there "high level
17 reports". For our purposes, I would wish to understand
18 what is comprehended by the expression "high level".
19 A. The content of high level reports would be a summary of
20 the number of outliers which would be outside the
21 parameters that were acceptable by the HES data; so
22 parameters were set of what was unacceptable and I think
23 Richard Willmer's example of a male hysterectomy would
24 be one of them, but there are a number of others. The
25 number of those incidents of unacceptable data within
0061
1 the parameters set can be and were in places fed back to
2 the coders so that you had, as it were, a postmortem of
3 the data submission and were able to be aware of some
4 maybe consistent problems that were always coming up and
5 that would enable you to talk to perhaps a particular
6 coder or a particular group of coders about some
7 inaccuracies. But when I say "high level", they were at
8 that very high level and would not have been about
9 reliability of individual diagnosis for individual
10 patients. But there was information that was helpful
11 and that could be used and it was not used consistently
12 across the country, it varied depending on individual
13 hospitals, feelings of the usefulness and the time that
14 needed to be spent on using those reports.
15 More recently, the hospitals have commissioned
16 from either other hospitals or from external agencies
17 external audit of their data, and the accreditation
18 process that we have set up and supported from the
19 information management group in its old guise is one
20 that will now be taken up, but there is some evidence
21 already existing through those accreditation studies of
22 the sort of feedback the sort of improvement that has
23 been made as a result of accreditation. So work has
24 been done in those areas.
25 I think we will probably find some information
0062
1 about that and make it available to the Inquiry, if that
2 would be helpful?
3 MR LANGSTAFF: Yes, please.
4 A. The next question is what variability is there in the
5 information that is available for coding? We touched on
6 that this morning and I made the point that some people
7 code from a discharge summary; others have available to
8 them all of the case notes. I think, again, practice
9 differed in different places. I think I would not be
10 able to comment except to say that variability of coding
11 is very dependent on having access to all the
12 information that is included in the case notes and that
13 the discharge summaries cannot possibly hope to give the
14 level of detail that would be needed to code very
15 accurately, unless you have very, very good discharge
16 summaries. So, again, the quality of the discharge
17 summary is going to make the coding either very good or
18 very difficult. I think looking at quality of discharge
19 summaries is a very important aspect to get an accurate
20 coding and I think having access to the full sets of
21 information available in case notes is the only way in
22 which a coder could really be sure that they were coding
23 completely accurately.
24 The next question was how important is clinical
25 coding taken in local organisations, and that is very
0063
1 variable. It is variable because different
2 organisations use and process information in a number of
3 different ways and the history has been that the
4 information that has been available has not been used as
5 fully as we could have used it in a number of different
6 areas.
7 We also have to remember that the information that
8 was mainly available through the data sets that we have
9 currently been looking at is information that has been
10 managerial and epidemiological, rather than directly
11 about individual patients and the outcomes for
12 individual patients.
13 So I think depending on the organisation,
14 depending on the skills within the organisation, the use
15 of data has been very variable and therefore the balance
16 of resources that have been put into using and analysing
17 that data has been very variable as well.
18 The last question is, if there is judgment in
19 coding, can it ever be accurate? I think the point that
20 I would want to make there is that the data that is
21 going to be used and analysed has to be fit for
22 a purpose, and one of the problems that we have had, and
23 still continue to have, is that we need to understand
24 when we collect data and we use data for what purposes
25 it is going to be used, so we can be sure that the
0064
1 definitions are used in processing and collecting those
2 data are standardised.
3 A lot of work has been done through the central
4 committee for regulating information requirements in
5 defining data definitions, and that is a process that
6 has been in place I think since 1991, where there were
7 standard definitions and standard processes defined for
8 the collection of data within Patient Administration
9 Systems, and therefore in subsequent submissions to the
10 HES data.
11 So those data definitions have been defined in
12 terms of the purpose for which that data was going to be
13 collected, and that has been for the managerial and
14 administrative processes rather than the clinical
15 processes and for its subsequent analysis for
16 epidemiological purposes.
17 So when there are differences of opinion about
18 what information could be used, what codes should be
19 used, I think they will be different because people will
20 have different purposes. I think it is being clear
21 about what purpose the information is collected for; can
22 it then be used subsequently for other purposes? This
23 is one of the things that affects people's judgment
24 about the use of codes and classification systems.
25 Those are the comments that I had to make on
0065
1 questions that were asked of Mr Willmer that I picked up
2 from yesterday.
3 Q. Thank you very much. You have supplied answers he was
4 not able to. There is nothing, apart from those few
5 matters I have raised with you in the course of your
6 evidence, which I want to ask you about that, because
7 essentially you are answering other questions, but it
8 may be that the Panel themselves have some questions for
9 you.
10 THE CHAIRMAN: There is just one question from
11 Mrs Howard.
12 Examined by THE PANEL:
13 MRS HOWARD: It is purely for the facts and for the record,
14 really. You talked about the organisation in Manchester
15 that trains coders. It would be useful if we had a name
16 for that organisation?
17 A. Can I give you the name of the organisation, when it
18 changed, but the current organisation? I can do that
19 with the dates?
20 MRS HOWARD: Thank you very much.
21 MR LANGSTAFF: May I thank you in your dual function as
22 expert commentator, and as giving evidence to us.
23 A. Thank you very much indeed.
24 THE CHAIRMAN: May I join in that thanks from the Panel's
25 point of view. You have helped us in two roles and we
0066
1 are very grateful to you.
2 (The witness withdrew)
3 MR LANGSTAFF: Dr Spiegelhalter, would you come forward,
4 please?
5 DR DAVID SPIEGELHALTER (RECALLED):
6 Examined by MR LANGSTAFF:
7 MR LANGSTAFF: I wonder if we can have back our slide
8 SLD 1/1 and turn it round --
9 THE CHAIRMAN: Mr Langstaff, for me and maybe others to
10 understand the capacity in which we are hearing
11 Dr Spiegelhalter, there is an oath to be administered or
12 not. Will you advise us, because others will be
13 interested in the answers.
14 MR LANGSTAFF: He gave evidence yesterday on oath. I think
15 you were sworn, were you not?
16 A. I affirmed.
17 Q. You affirmed when you first spoke to us. That will
18 cover anything that you have to say to us. You speak to
19 us as expert and adviser, and giving us the best of your
20 views as to the way forward now that we have, over the
21 past day and a half, had a public examination of some of
22 the evidence which bears upon the preliminary overview
23 of sources of data.
24 It will be appreciated by those who read this at
25 a distance that of course the oral evidence which we
0067
1 have heard is only part of the evidence which the
2 Inquiry has received, the rest of which was identified
3 yesterday evening and is in writing and which will
4 undoubtedly be supplemented by further written records
5 and evidence as and when they are received.
6 Indeed, if anyone has anything further which they
7 would wish to add, may I take this opportunity to
8 encourage them to give it to us at as early a stage as
9 possible so that you and other experts, and indeed the
10 Inquiry, can consider what they have to say.
11 We are now at the stage, really, of reviewing the
12 past day and a half and seeing whether we can move on
13 with any confidence at all, or at all, to Stages 2, 3
14 and 4 in the strategy which you mapped out for us
15 yesterday morning. Essentially, can we?
16 A. Perhaps if I could briefly review what we have covered
17 in Stage 1, the preliminary critical overview of sources
18 of data, we have heard about the HES system, the Cardiac
19 Register and the PAS system.
20 The HES system, the broad conclusions were that
21 data from 1991 onwards would seem of reasonable quality
22 to use. We would expect incomplete death notification,
23 due to it only containing deaths that occurred within
24 hospital, in general, and that of course there is
25 reliance on the accuracy of the coding.
0068
1 Perhaps the discussion this morning has encouraged
2 us as to the conscientiousness of the coding done in
3 Bristol in any way, for one example.
4 For the Cardiac Register, perhaps that has the
5 strong advantage of having clinical ownership of the
6 data, although it is non-validated and for both the HES
7 and Cardiac Register we have the problem that there is
8 no case mix available. However, those two sources of
9 data I think we should be quite encouraged about the
10 possibility of their use at least in the exploratory
11 stage to identify potential outliers.
12 In terms of the PAS, the clinical record system,
13 the PAS is useful both in informing us about the quality
14 of the HES data and as a source of information for
15 determining the clinical records in order to find out an
16 exact description of what has happened in Bristol over
17 the period of the Inquiry.
18 Of course, there are many other sources to find
19 the relevant clinical records and to cross-check these
20 that are listed in the documentation, for example
21 surgeons' logs and so on.
22 So from my personal perspective in Stage 1, what
23 we have heard in the last couple of days gives us some
24 good encouragement to carry on with the intended
25 analysis of the data that is available, with of course
0069
1 continual thought in our minds of the limitations of the
2 data -- the inevitable limitations of data, not being
3 collected for the purpose to which it is now being put.
4 None of this data was collected in order to allow this
5 type of examination.
6 So could I go on to the other stages?
7 Q. Yes, please.
8 A. The intention in Stage 2 is exploration. If we take
9 those three main sources of evidence, the HES data, the
10 Cardiac Register and the clinical records themselves,
11 then the intention is to analyse those three sources of
12 data independently and some of this has already started.
13 Q. That would be necessary because of the pressures of
14 time, that that should have begun?
15 A. Yes.
16 Q. Although obviously if the past day and a half had shown
17 we could not place any confidence in the process, it
18 would have to stop there?
19 A. Yes, but I think if there is any chance of getting this
20 completed, it would have to start as early as possible.
21 From the clinical records which are being recoded,
22 at least 1800 records over the period, that will provide
23 an accurate record and description of the experience of
24 the relevant procedures in Bristol over the period of
25 the Inquiry, which will be useful in order to determine
0070
1 exactly what happened in Bristol in terms of the
2 outcomes of the surgery.
3 Q. For those who may not have read the documents which are
4 on the Internet, records of each and every child who
5 underwent surgery in Bristol from 1984 to 1995, which is
6 presently available -- and we have heard this morning
7 that that is virtually all of them -- are being looked
8 at by a team of coders so that each and every case plays
9 a part in showing to us, independently of any other
10 source, what happened at Bristol.
11 A. Exactly, and that data, of course, can be presented in
12 various ways to look at aggregate mortality rates, for
13 example, over procedures and over certain periods of
14 time. That will provide us a very accurate description
15 of what actually went on.
16 The issue then becomes of comparing that with what
17 perhaps could be considered as standard practice,
18 standard performance in this country at that time for
19 the procedures under investigation.
20 So that is the use of the HES data and the Cardiac
21 Register.
22 The intention is that analyses of these will
23 proceed in parallel but independently, but as far as
24 possible -- and of course with the limitations -- to
25 a reasonably common protocol in that as far as possible
0071
1 to try to use common codes, as far as possible to use
2 a similar analysis and a similar presentation, in order
3 to help the eventual comparison of these sources.
4 Q. So to use the expression which I have used a couple of
5 times in questioning, to make sure that the data sources
6 are, for the purpose of comparison, "speaking the same
7 language"?
8 A. Yes, both in terms of the categorisations, but I think
9 it is also quite important that the presentations should
10 be similar as well, so it will enable someone looking at
11 the results to be able to reasonably rapidly compare
12 them. I think that is rather important.
13 Both of those sources take into account the
14 limitations of the data and inevitably involve some
15 pooling of the time. We should be able to examine what
16 was the overall standard according to those data sources
17 for these procedures at this period in this country, and
18 also -- this is I think very important -- to examine
19 between-unit variability in the outcomes. There is no
20 case mix adjustment in either of these sources so there
21 is inevitably going to be variability due to any number
22 of sources. It would be quite unreasonable to expect
23 all centres to be performing at exactly the same level,
24 even allowing for the chance fluctuations.
25 So to estimate that between-unit variability for
0072
1 the particular groupings of codes can be carried out;
2 and after doing that, and allowing for the chances due
3 to the low denominators inevitable in some of these rare
4 procedures, to identify, I will put it in quotes,
5 "outliers", centres who, for one reason or another,
6 appear to have a low or a high performance for the
7 conditions of interest and over the particular periods
8 being examined.
9 Q. If I can just stop you there, you mentioned the problem
10 with denominators and that is a function, I think, of
11 the small number of cases in respect of some
12 procedures. You mentioned earlier the need, I think, to
13 group.
14 In order to make sense of the data and what they
15 can show us, are we looking at the need to group
16 a number of procedures or conditions together if that
17 can appropriately be done?
18 A. It is a difficult problem. If you over-group, you can
19 lose the ability to identify what might be actually
20 rather important differences in some rather rare
21 interventions, but if you under-group, the denominators
22 are so low that variability can be largely explained by
23 chance alone. So there is an element of judgment there.
24 Q. So taking an extreme to illustrate the point: if all one
25 had to go on was looking at one particular record of one
0073
1 particular operation on the basis that every operation
2 on a human being is bound to be unique, because each
3 human being is unique to a greater or lesser extent, and
4 therefore if one looked at it at the level of the single
5 operation, one simply could not draw any conclusion,
6 positive or negative, as to the quality of surgery from
7 the fact of the outcome?
8 A. No, one operation could always be bad luck.
9 Q. Equally, if one takes an overall view, taking surgery as
10 a whole, if 10 or 12 per cent of people die, if that
11 were the case, that does not actually tell you anything
12 about particular procedures, particular disciplines,
13 particular approaches in particular units, so there has
14 to be some compromise between the individuality of every
15 operation and getting information which is actually
16 useful and tells you something you need to know on the
17 other hand. That is the point, is it?
18 A. Exactly. There are techniques for identifying outlying
19 institutions and in these two analyses carried out
20 independently, one should be able to then afterwards
21 compare whether they come up with broadly similar
22 conclusions, and if they do reinforce our confidence in
23 the conclusions or if they do not. There is an element
24 of judgment about which is perhaps the most reliable to
25 follow. Certainly yesterday we heard from Mr Keogh that
0074
1 from his professional perspective, he believed the
2 Cardiac Register more than the HES results.
3 In any case, the results that come out of there
4 can only be considered as exploratory at that stage
5 because of the inevitable limitations that we have
6 described.
7 Q. So at the end of that stage, we may have identified one
8 or more outliers, and you are were careful, I think, not
9 to say, "Well, we will identify whether Bristol is
10 different from other units", but "whether it is an
11 outlier". Perhaps you will give us some idea of how
12 different one has to be to be an outlier?
13 A. I think non-technically, to say somebody would be
14 considered quite a serious outlier if, taking into
15 account both the chance variability arising from the
16 small number of cases and taking into account the fact
17 that there is inevitably some between-unit variability
18 due to various unmeasured factors, they still had
19 performance that was extreme, taking into account those
20 two sources of variability, then they would label them
21 as a possible outlier for further investigation.
22 The next stage, then, really is trying to confirm
23 that "outlyingness" without actually still providing any
24 definitive explanations.
25 Q. Can I stop you there. Suppose that the exploratory
0075
1 stage shows no "outlyingness"?
2 A. There are still elements of the confirmation that should
3 take place, which I will come on to. But it would not
4 provide the perhaps focus of attention that is
5 essentially the purpose of that exploratory stage, the
6 point being that the exploratory stage, by throwing up
7 essentially hypotheses that one would like to go into in
8 more detail, permits the further analysis and further
9 investigation to focus on particular aspects.
10 So the next stage is to focus on what is thrown up
11 by Stage 2, and in particular the first thing would be
12 obviously to check Bristol's true experience, which the
13 clinical data will reveal in great detail, against what
14 has been reported or been analysed within the HES and
15 the Cardiac Register analysis. That enables us to
16 calibrate at least Bristol's contribution to those data
17 sources.
18 The problem comes, of course, in working out --
19 let us assume, as the evidence is so far suggesting,
20 that Bristol's data being submitted to those sources was
21 probably rather good -- essentially, whether we can
22 expect the other centres to have the same degree of
23 calibration between the true data reflected in the
24 clinical records and that being reported in the database
25 system.
0076
1 That is a difficult judgment to make. At that
2 point, there may be a need to consider other sources in
3 order to confirm within the areas under more detailed
4 investigation, what was the standard in the country at
5 that period and what was the background variability
6 between this and that over that period. I hasten to
7 add, this is very tentative, what we may need to
8 consider is perhaps other databases that other centres
9 have within these specific areas. It would be
10 unreasonable I think, at this stage to start trying to
11 explore every source of information on every case.
12 Q. So that is something for future review, and to be
13 reviewed no doubt in terms of feasibility and what one
14 would learn from it, and so on, in the light of what one
15 has derived from Stages 1 and 2?
16 A. Exactly, and I think in the light of perhaps
17 contributions from others on what they feel are perhaps
18 limitations on those data sources, people who might know
19 somewhat more about the quality of information provided
20 by other centres as well, to know whether those are
21 a reasonable benchmark or not. It is the benchmark that
22 we are looking for.
23 Q. So if, for instance, we had, in respect of a couple of
24 data sources from elsewhere, information which suggested
25 they were as reliable as Bristol may appear to be, that
0077
1 would give one more confidence in comparing Bristol with
2 the results of those units? In a lay understanding,
3 that is the point, is it?
4 A. Exactly. The final stage was, if we feel that at that
5 stage there still is some stronger evidence that Bristol
6 is being considered an outlier with respect to the
7 benchmark provided by the centres, we would seek
8 explanation for that. One possibility is to look at
9 more statistical explanation in terms of whether the
10 case mix being treated in Bristol differed
11 systematically from those in other centres. That again
12 may need to consider in the specific areas that have
13 been identified, collecting additional data from other
14 centres on their cases and their case mix.
15 I think at this stage, without prejudging the
16 issue, I would imagine only a reasonably rudimentary
17 adjustment for case mix would be possible, due to the
18 fact, as has been mentioned before in this particular
19 area, unlike adult cardiac surgery the development of
20 risk adjustment procedures based on more subtle
21 stratification has not been established.
22 In seeking to explain any variability that has
23 been to the best of my knowledge confirmed, it depends
24 of course on the explanation suggested by those with
25 a substantial knowledge of the area, but one study is in
0078
1 the process of being organised now which might
2 contribute to that explanation and in a sense, in order
3 for the evidence to be available in time when the
4 confirmatory analyses were being carried out, that study
5 really has to start taking place now in order to prevent
6 the evidence arriving really too late.
7 That is what one might call an "adequacy of care"
8 study, in which a stratified random sample of the 1800
9 or so clinical records that have already been identified
10 are taken and a multidisciplinary expert group assesses
11 that sample in terms of the adequacy of care delivered
12 to those children.
13 The form of that sample --
14 Q. Can I just stop you there. So you are saying that in
15 order to help look at the explanation, one needs to
16 have, from the 1800 or so cases that we have here in
17 Bristol, a lesser number drawn on, and you use the words
18 "randomised, stratified" -- I am sorry, I forget the
19 full words -- mix?
20 A. Stratified random sample.
21 Q. I was groping for the words, I am sorry. If you have
22 the stratified random sample, you will then be able to
23 make general conclusions from that sample?
24 A. To what one might consider to be a quantifiable extent,
25 because the purpose of statistical theory is to say to
0079
1 what extent you can generalise from a random sample of
2 a certain size.
3 Q. You are going on to describe how the stratified random
4 sample would work?
5 A. The idea is that it would be impossible to go through
6 all 1800 case notes to this level of detail required.
7 A simple random sample, if one just took a group at
8 random from those 1800, would not really reflect
9 adequately the cases of major interest in this Inquiry.
10 So the random sample is stratified so that cases that
11 are young and more severe have more chance of appearing
12 in the sample than those perhaps of simpler cases that
13 are not of such interest.
14 So that stratification scheme is preset, about
15 what defining what is meant by the young and more severe
16 cases and also it is not just the deaths but the
17 survivors that are being sampled to an equal number.
18 This enables an example that reflects the cases of
19 interest but also a broad range of types of patient
20 without specifically picking out known patients. These
21 patients are chosen completely at random, but weighted
22 in this way.
23 The initial sample will be of the size 80, and the
24 case notes will be subject to "blind review", the
25 reviewers will not know the identity of the patient, in
0080
1 terms of the adequacy of care which will be defined in
2 terms of levels of adequacy, distinguishing adequacy of
3 care, possible limitations of adequacy that would
4 possibly have led to a different conclusion or would
5 probably not have led to a different conclusion.
6 These will be distributed among the review panels
7 with some repeats going to multiple panels in order to
8 check the reproducibility of the assessments.
9 Q. The Inquiry has already published the names of the
10 experts who will advise it, getting on for 40 experts,
11 and they will be looking at 80 cases in order to draw
12 conclusions as you have indicated for the purposes that
13 you have set out.
14 A. 80 in the first wave; there may be more required.
15 Q. And you have mentioned there that there is an element of
16 cross-checking by random examples being given to more
17 than one group of four experts?
18 A. Yes.
19 The other aspect that should be mentioned,
20 although again, it is something that may need to be
21 considered, is that whereas in the mortality data we are
22 trying to benchmark the outcomes in Bristol by comparing
23 them with perhaps national standards, in terms of the
24 adequacy of care, it is rather difficult to establish
25 the national standard for what perhaps might have been
0081
1 the adequacy of care over that period for these
2 conditions. So it might be reasonable to explore the
3 possibility of providing a benchmark by considering
4 a limited exercise using the notes from other centres as
5 well.
6 Q. I have, I think, set out there the way forward that you
7 would expect in addressing what we see on our slide as
8 the explanation stage, at the end of which the
9 statistics, the available statistics, will have told us
10 what they can and enable such conclusions as can be
11 drawn from them to be drawn.
12 A. I feel that stage will not be largely a matter of
13 statistics.
14 Q. It will be largely a matter of analysis and
15 interpretation?
16 A. Yes.
17 MR LANGSTAFF: Thank you, Dr Spiegelhalter. May I say,
18 before you leave the witness stand, that what you have
19 said will undoubtedly be looked at carefully by others.
20 One of the purposes that the Inquiry have had in having,
21 today and yesterday, the public arena, as it were, was
22 to invite comment and discussion as to the way forward,
23 as I have indicated in my opening remarks, so that the
24 process could be refined and improved if that is
25 possible.
0082
1 Can I say, on behalf of the Inquiry -- I know the
2 Chairman is likely to echo this -- that we would welcome
3 any input from anyone who feels that he or she has
4 anything useful to add, whether by way of information,
5 whether by way of questions such as, "Have you thought
6 of this ...?", or comment, and it may well be experts in
7 their own fields who will have comments. Any of those
8 comments are invited because they need to be considered,
9 and considered carefully and over time, and of course
10 time in one sense we have something of before we
11 re-commence hearings in September, although, looked at
12 from another perspective, no great amount of time for
13 the investigations which have to take place.
14 So I would invite anyone who has comment, or
15 commentary, or advice or encouragement to give, to give
16 it sooner rather later if it is to be of the greatest
17 utility to us.
18 THE CHAIRMAN: And just interrupting for a moment, if
19 I may, I think it is a case, Mr Langstaff, that we will
20 be publishing a paper for consultation which sets out in
21 particular detail the sampling exercise that
22 Dr Spiegelhalter was referring to at the end of his
23 evidence a moment ago.
24 MR LANGSTAFF: Dr Spiegelhalter, unless there is anything
25 you would wish to add, may I thank you very much for
0083
1 your contribution and for being with us over the last
2 day and a half. I have no doubt that we will see you
3 again.
4 THE CHAIRMAN: And I, on behalf of the Panel, echo that
5 thanks. It must be clear by now, I think to all, how
6 complex are the issues that we are dealing with. It is
7 equally clear, I hope, that the scale of the enterprise
8 involved in seeking to resolve that complexity is huge,
9 but from the Panel's point of view, we see it as our
10 duty to undertake this work if we are fairly to draw any
11 conclusions which our terms of reference demand of us.
12 So we can only thank our experts who will carry out the
13 work and in particular, Dr Spiegelhalter, you today and
14 yesterday have begun us on the trail, as it were. We
15 owe you a great debt of gratitude, you and your
16 colleagues, not only for today but for your future
17 advice. Thank you.
18 (The witness withdrew)
19 MR LANGSTAFF: RE TIMETABLE
20 MR LANGSTAFF: Sir, if I can just review next week, as
21 I normally do towards the end of the session, next week,
22 on Monday and Tuesday, we recommence with Mr Wisheart,
23 whose evidence will go at this stage -- and I emphasise,
24 as I have done already -- to the question of
25 management. We have looked at the local scene. His
0084
1 evidence at this stage is part of the evidence that we
2 have been taking as to the management of the care. It
3 will not therefore deal with the adequacy of the care.
4 So the questions which we will be advancing to him
5 will not deal with any individual case; they will not
6 deal with the concerns that were expressed in whatever
7 time they were expressed, particularly perhaps in the
8 1990s, and they will not be concerned with the results
9 of any audit which was or was not carried out, and what
10 it did or might have shown.
11 I say that so that there is no misunderstanding as
12 to the scope of the questioning which Mr Wisheart will
13 face when he comes to see us and give his evidence on
14 Monday and Tuesday of next week.
15 On Wednesday and Thursday, the last two days that
16 we have before a break in the summer, we will sit to
17 listen to Professor Green of the Royal College of
18 Pathologists. They have recently published
19 a consultation paper which deals with, of particular
20 interest to us, the question of consents for postmortem
21 and retention of tissue. Mr Clifford then, from the
22 Home Office Coroner's Unit.
23 On Thursday, Mr Burgess of the Coroners' Society
24 and Diane Kennington, the Patient Affairs Officer of the
25 UBHT.
0085
1 Those four witnesses will revisit the issue of the
2 retention of hearts which we began this week with, and
3 they will help, we hope, to set both the national and to
4 some extent the local background scene against which the
5 particular evidence of what happened particularly in
6 this case in September will then be viewed. As
7 I indicated at the start of this week, in September we
8 will hear from Professor Berry and Dr Ashworth, and of
9 course from parents and their reaction as to what
10 happened to them in respect of what had happened to
11 their children whilst under the care of the Bristol
12 hospitals.
13 Sir, on Monday we begin at 10.30. On each of the
14 other days next week, at 9.30.
15 Sir, that would normally conclude the business for
16 this week, but before we close, Mr Lissack has an
17 application to make. I think it relates to timetabling
18 for next week.
19 It is perhaps convenient if I give him my
20 microphone, given the way in which, because of the
21 exigencies of the evidence, the hearing chamber has been
22 configured.
23 MR LISSACK: RE TIMETABLE
24 MR LISSACK: Sir, as you are aware, it has been our practice
25 throughout the Inquiry to keep Mr Langstaff and, through
0086
1 him, you informed of any problems which we foresee which
2 may cause a dislocation of the programme that you have
3 set.
4 We hope that that early warning system and the
5 constantly open lines of communication that we have with
6 Mr Langstaff has assisted in the smooth running of the
7 Inquiry to date.
8 It is in that spirit that I was minded to make an
9 application to you today. Happily, the application that
10 I was envisaging having to make, I need not, because an
11 accommodation, the details of which I need not go into,
12 has been reached which I hope may provide that a formula
13 by consensus to ensure that the programme in the early
14 part of the autumn session is not disrupted, is reached.
15 However, the matter I do apply to you about is
16 this: it is an application for an application, really.
17 I have discussed this with my learned friend
18 Mr Langstaff and we have agreed this is an appropriate
19 way of presenting the issue.
20 Next week is congested. There is a lot of
21 important evidence to be heard. But we would invite
22 you, please, to allow, in the timetable, for the need
23 for us to apply to you for an order to compel a party or
24 parties to produce material.
25 Our concern is simply this: if the consensus
0087
1 approach which we are proposing to adopt with
2 Mr Langstaff's assistance -- I am grateful to him for
3 it -- does not work, then there is a real risk that if
4 the Inquiry breaks without more, that when we revisit
5 the issue of organ and tissue retention in September, we
6 will actually be little forward from where we are now.
7 That is highly undesirable for the Inquiry, we submit,
8 and it is the last thing we wish to happen.
9 So beyond that, I need not go.
10 Whilst I am here, may I please say two other
11 things, which it would be remiss of me to fail to
12 mention.
13 Firstly this: my clients, and if I may say so,
14 with them, myself and the others who act on behalf of
15 the BHCAG, are very grateful to the Inquiry for
16 dislocating your timetable, as we know you have, at this
17 difficult time in the scheduling. They would ask me to
18 thank you expressly not only for the rescheduling that
19 you have permitted, but also for the sensitivity that
20 the Inquiry has shown in the handling of the issue of
21 retention of tissue, both inside and outside the
22 chamber.
23 Secondly, I am instructed, and welcome the
24 opportunity to pay tribute to those responsible for the
25 collating and presentation of the statistical evidence.
0088
1 We immediately saw the importance of this material
2 and evidence which we have paid close attention to
3 during its presentation and in the reading of the
4 material beforehand, because we respectfully agree that
5 it is key to forming a proper conclusion and proper
6 evaluation in a fair and balanced way of how Bristol
7 performed, to understand how it performed in the context
8 of other similar units carrying out the same work.
9 We hope that the evidence that has been provided
10 to date and will be elaborated upon in due course, will
11 be of considerable assistance in weighing the
12 significance of Block 6 evidence which of course closely
13 concerns our lay clients, and ultimately will be of
14 considerable assistance to you in the shaping of your
15 recommendations as to how data may be handled in the
16 future, a matter upon which we will be making written
17 submissions in due course.
18 But for the way in which it has been marshalled
19 and presented, speaking for myself, it would have been
20 a case of the uncomprehending in pursuit of the
21 incomprehensible!
22 Thank you very much indeed.
23 THE CHAIRMAN: Mr Lissack, thank you very much. I could
24 not claim to adopt the last of those observations; it is
25 clearly wrong. But thank you first for the kind remarks
0089
1 you have addressed both to the Panel and to those who
2 have been assisting us. We are very grateful. It is
3 always nice to hear nice things said, because very often
4 in this position, not so nice things are said, so we are
5 grateful and we shall store it away against harsher
6 times!
7 Lastly, and on the substantive point you raise,
8 I am very grateful to you for having pursued, as I know
9 assiduously, a means whereby you can come to an
10 agreement with others, and I am pleased to hear that so
11 far, at least, some consensus appears, but you rightly
12 say, and I agree, that there ought to be a "Plan B", if
13 I can put it that way, so of course should it be
14 required, time will be made available to you, yes, I can
15 give you that assurance.
16 MR LISSACK: Thank you.
17 THE CHAIRMAN: I will go on to say, for the general public
18 and for you, although we will be sitting from Thursday
19 of next week onwards, I am of course available, as are
20 those who advise me, throughout the whole of the
21 summer. Should matters arise upon which you need my
22 intervention or the intervention of anyone else, I am at
23 your disposal and always will be. I give you that
24 assurance as regards time, should it arise.
25 MR LISSACK: Thank you very much indeed.
0090
1 THE CHAIRMAN: Mr Langstaff?
2 MR LANGSTAFF: Sir, that concludes today's, and indeed this
3 week's, open business.
4 THE CHAIRMAN: Thank you very much indeed, everyone.
5 Thank you, Mr Langstaff.
6 We reconvene on Monday next at 10.30.
7 (12.56 pm)
8 (Adjourned until Monday 19th July 1999 at 10.30 am)
9
10
11
12 I N D E X
13
14
15 MR ANDREW HOOPER (affirmed)
16 Examined by MR LANGSTAFF .......... 1
17 Examined by THE PANEL .......... 49
18 MISS ANN HARDING (affirmed)
19 Examined by MR LANGSTAFF .......... 56
20 Examined by THE PANEL .......... 66
21 DR DAVID SPIEGELHALTER (recalled) .......... 67
22 Examined by MR LANGSTAFF .......... 67
23 MR LANGSTAFF: RE TIMETABLE .......... 84
24 MR LISSACK: RE TIMETABLE .......... 86
25
0091
Return to top of page
Published by the Bristol Royal Infirmary Inquiry, July 2001
© Crown Copyright 2001