Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Unreliability of volunteered mortality data
- John L Gibbs, David Cunningham (13 February 2002)
-
Correction - letter of Feb 13th, Gibbs & Cunningham
- John L Gibbs, David Cunningham (5 March 2002)
-
Re: Unreliability of volunteered mortality data
- Laurence N Vick, Gabriel Rogers (20 March 2002)
-
Re: “BRI – Mortality and volume of cases in paediatric cardiac surgery
- James P Stewart (12 June 2002)
|
|
|||
|
John L Gibbs, Lead Clinician, UK Congenital Heart Disease Central Cardiac Audit Database Leeds General Infirmary, Great George St, Leeds LS1 3EX, David Cunningham
Send response to journal:
|
Editor – David Spiegelhalter’s article1 adds to the already persuasive data from the early 1990’s suggesting an inverse relationship between volume of cases and mortality in congenital heart disease surgery. But he goes on to compare mortality derived from the Cardiac Surgical Register (CSR), Hospital Episode Statistics (HES) and the Bristol Inquiry. This comparison is fundamentally flawed and is almost certainly unfairly biased against Bristol. Mortality rates in Bristol have been so closely scrutinised for the period covered by the inquiry that they are highly likely to be accurate. There has, however, to our knowledge been no attempt to validate the volunteered mortality data from CSR or HES data from the same period. The UK Central Cardiac Audit Database (CCAD),has collected data from all UK congenital heart disease centres since April 2000, including volunteered mortality data. CCAD, in contrast to the CSR and HES, also tracks mortality independently using the patient’s NHS number and a direct link to the Office of National Statistics. Volunteered and centrally tracked 30 day mortality rates differ considerably. Overall tracked mortalities were 25% higher than reported mortalities. 7 of 11 centres in England under-reported early mortality, sometimes because patients were discharged very early, but also sometimes because the reporting was erroneous. Six patients who died within 7 days of operation were wrongly reported as alive at discharge. Furthermore, use of Hospital Episode Statistics did not improve accuracy of status reporting. In a sample of nearly 3,000 procedures carried out in between 1/4/2000 and 31/3/2001, HES data under-reported the total number of procedures by 10% and under-reported 30 day deaths by 9%, but also classified 1% of surviving patients erroneously as deceased. We understand that links between ONS and HES are being explored but those links were not in existence when the HES 2000-2001 data became available or at the time of the Bristol Inquiry. It is likely that the differences between mortality in the volunteered CSR data and the Bristol Inquiry data were of at least equal magnitude to those described above. Any new or past comparison of mortality rates which fails to take into account the difference in data quality from non validated sources and from the Bristol over the period of the Inquiry risks doing serious injustice to Bristol as well as to the profession’s ongoing attempts to restore the public’s confidence in congenital heart disease services in the UK. John L Gibbs FRCP
David Cunningham PhD
1. Spiegelhalter D. Mortality and volume of cases in paediatric cardiac surgery: retrospective study based on |
|||
|
|
|||
|
John L Gibbs, Consultant cardiologist Leeds General Infirmary LS1 3EX, David Cunningham
Send response to journal:
|
Unreliability of volunteered mortality data John L Gibbs, et al. bmj.com, 13 Feb 2002 Editor – on re reading our letter of 13th Feb in response to Spiegelhalter's article of case volume and mortality we are concerned that our statement "Overall tracked mortalities were 25% higher than reported mortalities" may have given the impression that total tracked mortality in the UK was 25% higher than reported mortality. What we should have said is "Overall tracked 30-day mortality in individual hospitals was up to 25% higher than reported discharge mortality". Our message concerning the dangers of comparison of validated and unvalidated data remains unchanged. John L Gibbs
|
|||
|
|
|||
|
Laurence N Vick, solicitor Michelmores, EX1 1HE, Gabriel Rogers
Send response to journal:
|
Editor – The response submitted by Drs Gibbs and Cunningham [1] to Dr Spiegelhalter’s article presenting data collected during the Bristol Royal Infirmary Inquiry [2] presents a cogent argument, but is based on a false premise. Dr Spiegelhalter is not guilty of using some sort of validated and/or synthesised data source to provide mortality rates for the Bristol paediatric cardiac unit for the period in question: as with the other units, the Bristol figures are merely those that were volunteered to the CSR or HES by the clinicians concerned at the time in question. This has the disadvantage that such data is unverified and potentially unreliable but, at least, thereby puts Bristol on a “level playing field” with other units.
It is interesting that Drs Gibbs and Cunningham assume that an authoritative record of paediatric cardiac surgery at Bristol would be available; it would, indeed, have been fairly straightforward (if somewhat laborious) for the Inquiry to have put together such a resource. However, although the statisticians whom the Inquiry engaged devoted a certain amount of effort to analysing the extent to which the different sources of mortality data “agree”, no attempt was made to synthesise the sources in question, to cross-reference each to the other in an attempt to produce the kind of accurate figures which Drs Gibbs and Cunningham postulate. One would assume that the Bristol Royal Infirmary Inquiry – somewhere amongst its 14 million pounds’ worth of work spread over 3 years – would be able to answer the questions, “How many children received heart surgery at Bristol, and how many of these died?” These questions remain unanswerable. It is to be hoped that the UK Central Cardiac Audit Database will not only greatly ease the generation of reliable and truly comparable mortality rates, but will also provide the kind of “early warning system” that will prevent future tragedies on the scale of Bristol. I note with extreme regret, however, that CCAD shows no inclination to track non-fatal outcomes to cardiac surgery. Children who survive heart surgery with brain damage or other non-fatal complications have been – and remain – grossly overlooked in the analysis of cardiac surgical performance (for example, this issue receives precisely one paragraph in the Kennedy report). People are often surprised to learn that there is absolutely no way of establishing whether the incidence of brain damage following heart surgery was any higher at Bristol than elsewhere; they are astonished to learn that there is still no way of comparing various centres’ records in this area. Dr Kate Bull has produced an extremely constructive paper on this subject [3], and the Brompton and Harefield Inquiries made equally useful observations and recommendations [4]. Unfortunately, those who commission such views in the first place have persisted in all-but ignoring the findings (whilst, of course, offering lip-service to the contrary). Laurence Vick Solicitor, Michelmores, 18 Cathedral Yard, Exeter. EX1 1HE. e-mail: lnv@michelmores.com 1. Gibbs JL, Cunningham D. Unreliability of volunteered mortality data. bmj.com, 13 Feb 2002. 2. Spiegelhalter D. Mortality and volume of cases in paediatric cardiac surgery: retrospective study based on routinely collected data. BMJ 2001;323:1–5. 3. Bull, C. Key Issues in Retrospective Evaluation of Morbidity Outcomes Following Paediatric Cardiac Surgery. Bristol Royal Infirmary Inquiry, 2000. [available online at http://www.bristol-inquiry.org.uk/images/seminars/Bmr.pdf] 4. Evans R et al. Report of the Independent Inquiries into Paediatric Cardiac Services at the Royal Brompton Hospital and Harefield Hospital. 2001. [available online at http://www.rbh.nthames.nhs.uk/GENERAL/Press/Paediatricinquiry/FullReport.doc] |
|||
|
|
|||
|
James P Stewart
Send response to journal:
|
Dear Sir, Re: “BRI – Mortality and volume of cases in paediatric cardiac surgery” Mr Spiegelhalter’s paper presents further confirmation that the BRI’s mortality record was extremely poor. What this paper does not answer is why the mortality record was so poor and what was the BRI’s true ‘success’ rate. Many children who survived with appalling brain damage are counted as ‘successes’. As the father of a child who suffered horrendous brain damage, I find it outrageous that Mr Spiegelhalter should count him as a ‘success’. It is long past time that an in depth study was conducted of what were Bristol’s TRUE success rates and WHY did so many children die or suffer morbidity. The answer is largely to be found in the fact that the surgeons were very slow. Their operation times, times on bypass and cross clamp times were all excessive. As Mr Pawade said at the GMC, ‘the fate of the child is decided in theatre’ and ‘it is a race against time’ . Similarly, Professor Angellini concerning morbidity said, ‘complications are a well known fact to be enhanced by prolonged operations’ In respect of bypass times, a Great Ormond Street Hospital study says that patients with adverse neurological events had a significantly longer mean cardiopulmonary bypass time than normal survivors (113 v 93 minutes; p<0.05), as did patients who subsequently died (199 v 93 minutes; p<0.001). If one examines Mr Wisheart’s operations, taking 113 minutes bypass time as per the above study as an indicator that the length of time on bypass has probably (p<0.05) resulted in an adverse ‘neurological event’. One finds that from a sample of just over one hundred operations that approximately 60% of their bypass times exceeded 113 minutes. The horrifying implication of this, is that the majority of Mr Wisheart’s operations suffered brain damage. The question of how many children survived with brain damage has never been examined. The recent £14 million pound Public Inquiry carefully avoided this issue. No doubt due to the fact that the solicitors who ran this Inquiry were from the Treasury and that they had a vested interest in ensuring that these sorts of studies and facts were avoided. It also explains why my wife and I were not allowed representation and why we were effectively barred from this supposedly ‘Public’ Inquiry. It is in the Public Interest that the true failure rate at the BRI should be ascertained. The reasons for this need to fully revealed. I urge Mr Spiegelhalter to undertake a comprehensive study that focuses on the surgeons’ surgical times, their bypass times, their cross-clamp times, the times to extubation and the statistical association that this has on excess mortality and morbidity at the BRI. Far too much time and money has already been wasted by the Public Inquiry on such spurious reasons as the ‘split-site’, which played a negligible role as evidenced by the fact that Mr Pawade used exactly the same facilities for his first 88 operations with only one death . Why was Mr Pawade’s mortality rate some fifteen times lower than Mr Wisheart’s? Could it be because he had been properly trained, because he recognised that ‘speed is of the essence’ and that the child’s fate is largely decided in theatre? Yours sincerely Jim Stewart
1) GMC 9-5F and 9-6D respectively. 2) GMC3-63B. 3) Incidence of neurological complications of surgery for congenital heart disease by P Fallon, JM Aparicio, MJ Elliot and FJ Kirkham – Archives of Disease in Childhood - 1995 4) As contained in the internal audit conducted by Dr Bolsin and Dr Black 5) GMC 9-1C |
|||