Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
The importance of data quality
- David R Walker (4 July 2003)
-
True performance assessment requires assessment of high-risk cases
- Michael J O'Leary (7 July 2003)
-
Logistic EuroSCORE to assess cardiac surgical mortality
- Jonathan I Ferguson, Andrew J Drain, Sharon Wilkinson, Samer Nashef (17 July 2003)
-
Zero tolerance for deaths and complications in cardiac surgery
- Richard G Fiddian-Green (18 July 2003)
-
Crude and stratified data reporting: the weakest link
- John M Alvarez (21 July 2003)
-
Comparison of low risk cases is far from ideal
- Bhaskar Kumar, Dilip Oswal (2 September 2003)
-
Quality and outcomes studies in Italy
- Emanuela Taioli, Alessandro Ghirardini, Donato Greco, and Alessandro Nanni Costa (2 September 2003)
|
|
|||
|
David R Walker, Director of Public Health County Durham and Tees Valley Strategic Health Authority
Send response to journal:
|
I would fully support the use of risk stratified mortality data in preference to crude mortality data if surgeon specific comparisons are to be published. However. the quality of such information depends upon the rigour with which data is collected. In this well-conducted study, great care was taken to ensure completeness and accuracy of data, including the prospective collection of data, such that only 2% of data items were missing. This adds considerable weight to the conclusions reached. If such methods are to be used nationally, there must be equally robust validation of the data collection process in all institutions to prevent a potential information bias. There is a danger that retrospective data collection will be used, and we know that the investment in, and quality of, coding and record keeping varies markedly between institutions. The absence of data concerning co-morbidities or the failure to code them, for example, will result in a low estimate of expected mortality and will falsely suggest a poorer surgical performance. I would suggest that, should this approach be adopted, robust evidence of data completeness and accuracy should be a prerequisite before data is accepted from any institution to ensure that we are comparing surgical performance and not administrative capacity. Competing interests: None declared |
|||
|
|
|||
|
Michael J O'Leary, Specialist in Intensive Care Sydney 2217 Australia
Send response to journal:
|
Risk-adjusted outcomes are essential for the interpretation of surgeon specific mortality, however, Bridgewater and colleagues inform us such adjustments do not discriminate in low-risk patients (the majority) where baseline mortality is very low. They claim that the differences in surgeons' crude mortality rates are explained by variation in the case-mix of high-risk patients which is not amenable to risk stratification. However it is probably only within this group that true differences in performance will be detectable. It is precisely within this group of patients that an under performing surgeon is likely to create most havoc! Bridgewater et al recommend "a comparative analysis based on low risk cases without the need for further risk adjustment". All this will achieve is a demonstration that all surgeons fall within two standard deviations of the mean for the average case. Take another pat on the back! It would be interesting to know the individual surgeons' "risk- adjusted" performance in the high-risk cases as it may well be that this would mirror their performance in the crude, non-adjusted rates and consequently that crude mortality rates are actually the most accurate way to procede. Competing interests: None declared |
|||
|
|
|||
|
Jonathan I Ferguson, SpR Cardiothoracic Surgery Papworth Hospital CB3 8RE, Andrew J Drain, Sharon Wilkinson, Samer Nashef
Send response to journal:
|
Editor- We congratulate Bridgewater and colleagues 1 on a well- constructed audit of mortality and risk assessment in adult cardiac surgery. Risk stratification is crucial in surgical decision making, informed consent and assessing surgical performance. Publication of raw (non-risk-stratified) surgeon-specific mortality rates may lead some surgeons to practise risk-adverse behaviour by not offering surgery to high-risk patients who have often the most to gain from cardiac surgery. Those surgeons who do operate on high-risk patients may find themselves unfairly vilified. Bridgewater’s concern about the potential damage to patients and surgeons is real. It is a pity that the study did not use the logistic EuroSCORE. The authors state that the logistic model is neither widely used nor validated. We disagree. Reports to the EuroSCORE project group indicate wide use throughout Europe and elsewhere. As for validation, it has to be remembered that the additive model is only an approximation of the logistic model. If the former is validated fully, the latter can only be more so! The authors’ concern about the additive model underscoring in high- risk patients has led them to recommend that only low-risk patients should be used to assess surgical performance. Logistic EuroSCORE is accurate for high and low risk patients 2 and subsequently all patients can be used to assess surgical performance. Studying only low-risk patients in assessing performance poses a practical statistical problem. Few deaths will occur in this group, giving low mortality rates with wide confidence intervals. Take a hypothetical surgeon with an actual mortality of 2% in a group of patients with a predicted mortality of 1% (twice expected). It would take 600 cases in this risk band for the surgeon’s results to be significantly worse than predicted (with 95% confidence). At a realistic annual surgical throughput of 50 low-risk cases it would take 12 years and 6 low- risk deaths for the underperformance to be detected! Analyses which include all patients (low and high risk), properly stratified using models like the logistic EuroSCORE, would obviate this problem. The standard or additive EuroSCORE represents a great advance in allowing risk prediction at the bedside. Risk factor combinations, however, are not additive and their combined effect is greater than the sum of their parts. Although the additive model remains suitable for large patient populations and is attractive to units embarking on risk study, sophisticated hospitals with ready access to information technology (like those in the Northwest Quality Improvement Programme) should consider the logistic model. It requires no additional data collection and the risk calculation can be carried out using one of the many free-to-download or online calculators (www.euroscore.org). 1. Bridgewater B, Grayson A D, Jackson M, Brooks N, Grotte GJ, Keenan DJM, Millner R, Fabri B M, Jones M on behalf of the North West Quality Improvement Programme in Cardiac Interventions. Surgeon specific mortality in adult cardiac surgery:comparison between crude and risk stratified data. BMJ 2003;327:13-17. (5 July.) 2. Michel P, Roques F, Nashef SAM. The EuroSCORE Project Group. Logistic or additive EuroSCORE for high-risk patients? European Journal of Cardio-thoracic Surgery 2003; 23: 684–687 Competing interests: S A M Nashef is leader of the EuroSCORE project |
|||
|
|
|||
|
Richard G Fiddian-Green, None None
Send response to journal:
|
The residual risk of coronary bypass grafting, which has become refined and is performed with low risk on large numbers of patients by most cardiac surgeons, lies in the details. As the age and number of comorbidities in the surgical population presenting for coronary artery bypass grafting increases, strategies to lessen operative risk have been devised. Off-pump bypass grafting to avoid the detrimental effects of extracorporeal circulation is the latest innovation. It has not been without its complications and may have been introduced prematurely and even inappropriately. It may have been more appropriate to have first revisited the benefits and risks of pulsatile relative to non-pulsatile perfusion during cardiopulmonary byass. Recent data suggest that off-pump bypass grafting might have increased the risks of cardiac surgery. This is not surprising for beating heart surgery exposes the myocardium to a continued workload during the coronary vascular occlusion needed to construct coronary bypass grafts. This continued workload is greatly diminished by stopping the heart from beating during cardiopulmonary bypass. The potential for myocardial demand for energy from ATP hydrolysis to outstrip the capacity for resynthesis during beating heart surgery can be expected, therefore, to be much greater than that during cardiopulmonary bypass when the heart is not beating. Avoidance of the cardiopulmonary bypass circuit has been linked to the development of a hypercoagulable state postoperatively. Other complications related to the unique management of the ascending aorta and target vessels during the performance of beating heart surgery have also been reported. More importantly, despite increasing experience in a number of centers, haemodynamic collapse has occured during off-pump bypass, necessitating the rapid institution of cardiopulmonary bypass (1). The haemodynamic collapse might have been the product of an excessively large myocardial or systemic energy defict. A myocardial energy deficit might have developed during the coronary artery occlusion required for the construction of the grafts. A systemic energy defict might have developed at the same time from an impairment of myocardial performance and hence systemic O2 transport whilst the coronary arteries were occluded. In a recent study performed by cardiac anaesthetists at Columbia University College of Physicians and Surgeons, New York, 19 patients who had insertion of a left ventricular assist device were enrolled in a prospective, observational study (2). Compared with baseline, systemic blood flow and by inference O2 transport was significantly increased at the end of operation (p < 0.0001). Tonometric variables, which were normal at baseline, became abnormal in 90% of patients (baseline CO2 gap 4 +/- 2 mm Hg versus end of operation CO2 gap 24 +/- 15 mm Hg, p < 0.0001). Elevated CO2 gaps correlated with larger doses of norepinephrine (r = 0.69, p = 0.001) and vasopressin (r = 0.88, p < 0.0001). Abnormal gastric tonometric variables at the end of operation correlated with postoperative intensive care unit length of stay (r = 0.70, p = 0.0009) and multiple organ dysfunction score (r = 0.64, p = 0.0033). These findings are consistent with ours performed in 85 patients having mostly bypass grafting (3). These and many other supporting studies establish that the pCO2-gap, pH-gap and gastric intramucosal pH, different tonometric indices of tissue bioenergetics, are the most important monitored determinants of outcome after cardiac surgery as expected from the pathophysiology. In another recent study, performed in Southampton, 54 patients undergoing primary coronary artery bypass grafting by the same surgeon were randomized into either on-pump of off-pump groups. In both groups there was a similar and progressive drop in gastric pHi intraoperatively. Postoperatively, there was a gradual separation between the groups with the on cardiopulmonary bypass patients showing no further decline in pHi, while further deterioration was observed in the beating heart group up to 6 h postoperatively. There was a significant difference between the groups over time (P=0.03) (4). This study establishes that beating heart surgery has a detremental effect upon splanchnic tissue bioenergetics and hence potentially upon outcome. Pulsatile perfusion was employed during cardiopulmopnary bypass in the Southampton study in preference to non-pulsalatile perfusion. This is an important difference for, as Professor Ken Taylor at the Hammersmith has shown, non-pulsatile perfusion reduces the risk of cardiopulmonary bypass and in a restrospective analysis of a large number of his cases improved outcome as expected from the pathophysiology (5). These data are compelling grounds for halting beating heart surgery except in the context of properly designed and monitored prospective studies comparing it with conventional surgery. Given the data it would seem, however, most appropriate to compare it with conventional surgery performed during pulsatile perfusion rather than during non-pulsatile perfusion. In which case a prospective randomised study comparing pulsatile with non-pulsatile perfusion should precede it to establish whether outcome is indeed better with pulstalile perfusion than with non- pulsatile perfusion as would seem likely from the pathophysiology and as Professor Taylor has advocated for many years. Many companies have a financial interest in maintaining the status quo and in the evolution of minimally invasive beating heart surgery. Minimally invasive programs attract patients and are good for hospital business in the US where cardiac programs are often the cash-cow supporting other departments if not the entire hospital. Because so many surgeons in so many centers are now performing these operations with equivalent results profit margins in these highly efficient assemply lines have fallen. Most hospitals, certainly in the US, have invested large amounts of money in their cardiac programs and are reluctant to make any new investment in the program unless it can be shown to cut costs or increase revenues. The standard argument has been that the mortality and morbidity of cardiac surgery are now so low that it is not cost-effective to change over to pulsatile perfusion. That argument assumes that some deaths and complications are an inevitable part of doing surgery. That argument is no longer tenable. Zero tolerance for post operative deaths and complications has to become the standard even in octogenarians with co-morbidities. That is not unreasonable for all deaths and complications appear to be the product of shock, defined as an abnormally low gastric pHi, co-morbidity simply reducing the tolerance for shock. Had zero tolerance been the standard in the first place beating heart surgery may never have been introduced into clinical practice in the US. 1. Dewey TM, Edgerton JR. Complications related to off-pump bypass grafting. Semin Thorac Cardiovasc Surg. 2003 Jan;15(1):63-70. 2. O'Malley CM, Frumento RJ, Mets B, Naka Y, Bennett-Guerrero E. Abnormal gastric tonometric variables and vasoconstrictor use after left ventricular assist device insertion. Ann Thorac Surg. 2003 Jun;75(6):1886-91. 3. Fiddian-Green RG, Baker S. Predictive value of the stomach wall pH for complications after cardiac operations: comparison with other monitoring. Crit Care Med. 1987 Feb;15(2):153-6. 4. Velissaris T, Tang A, Murray M, El-Minshawy A, Hett D, Ohri S. A prospective randomized study to evaluate splanchnic hypoxia during beating -heart and conventional coronary revascularization. Eur J Cardiothorac Surg. 2003 Jun;23(6):917-24. 5. Fiddian-Green RG. Gut mucosal ischemia during cardiac surgery. Semin Thorac Cardiovasc Surg. 1990 Oct;2(4):389-99. Review Competing interests: None declared |
|||
|
|
|||
|
John M Alvarez, Consultant Perth, W.Australia 6009
Send response to journal:
|
Bridgewater and colleagues are to be congratulated on unequivocally evincing that coronary bypass surgery, particularly in low risk cases, is performed with very good results across a diverse land mass (i.e. Northern England) and not just in a highly specialized international center of excellence1. Importantly, half of all deaths occurred in high risk cases and although a strong association was found between volume and outcomes no further comment was possible given the study design. One of the motivating factors for such a study is that surgeons will, and many already do, develop risk averse behaviour (i.e. avoid the “tough cases”) despite these patients being those who stand to gain the most from surgery, because of a fear by the surgeon of litigation or suspension in the event of an adverse event. A profound weakness of papers of this genre is the absence of details stating just what did these patients die of and when did this causative event occur: during anaesthetic induction, actual surgery, perfusion management or postoperative intensive or non intensive care management? Unquestionably, in many cases, but by no means all, no specific adverse misadventure may be identified. The absence of such details and their analyses may explain the poor correlation of scoring systems between high and low risk cases and case volumes and outcomes. Failure to recognize signs of ischemia on induction will have a profound effect on outcome if an infarct develops and the surgical strategy proceeds unaware and unchanged, progressive inotrope use inducing organ shutdown needing dialysis from inadequate recognition of either pneumothorax, intravascular hypovolaemia or atrial fibrillation are likely to be malignant in outcomes. These are not uncommon witnessed examples which regardless of whether the case was high or low risk will have a profound impact independent of the quality of the operative surgery. So too a massive anterior infarct from a single graft onto a good sized left anterior descending coronary artery will unlikely be influenced by the quality of the anesthesia or intensive care. Surgical mortality tables reflect on all the team players. Ostensibly therefore excellent articles as this one is, need to address these specific issues if clinical governance is to be effective; yet, as is often, the silence remains deafening. The strength of a chain is not measured by the strongest link rather by its weakest component. Yours sincerely Mr. John M. Alvarez FRACS
1.Bridgewater B,Grayson AD,Jackson M,Brooks N, Grotte G,J,Keenan DJM et al. Surgeon specific mortality in adult cardiac surgery: comparison between crude and risk stratified data. BMJ 2003;327:13-17 Competing interests: None declared |
|||
|
|
|||
|
Bhaskar Kumar, Research Fellow Leeds general Infirmary, LS1 3EX, Dilip Oswal
Send response to journal:
|
Editor – The article by Bridgewater et al[1] compares the difference between crude mortality and risk stratified results for performance of cardiac surgeons. Their principal finding was that surgeons cannot be accurately compared on the basis of surgeon specific, crude mortality data. They propose that comparative analysis should be based on low risk cases. However, we feel that two vital issues will be overlooked if this is implemented. The use of low risk cases as a benchmark comparison as suggested by the authors will in fact promote aversion behaviour amongst consultants in attempts to keep their mortality low. This would lead to low risk cases being taken away from trainees. This would further reduce trainees surgical exposure which is already compromised by the European Working Time Directive. Therefore it is important that low risk cases should not be kept solely for outcome monitoring purposes and should be maintained as a vital ingredient for the training of future generations of cardiac surgeons. Secondly, placing the burden of accountability entirely on the surgeon masks the influence of a variety of factors which account for the final ‘surgical outcome’. In the context of cardiac surgery this includes the timing of referral by a cardiologist, accuracy of investigations, optimisation of medical therapy before the operation especially in acute cases, unpredictable delays in surgery due to lack of intensive care beds and the quality of peri-operative care provided by anaesthetists, intensivists, pefusionists and the nursing staff. The availability of resources for monitoring and therapeutic modalities must also be taken into consideration. If a patient dies from an unrelated event like ischaemic bowel, it is still included in a surgeon’s operative mortality. General public and media reading the individual surgeon’s results will never understand all these variables. For them interpretation will be very simple – a surgeon with mortality of 2.0% is a better surgeon than the one with a mortality of 2.5%. This interpretation can be very dangerous not only for the surgeons but for the patients and the National Health Service as a whole. We call for the use of a results monitoring system which spreads the burden of responsibility towards all of those involved in the patient care. Comparison of low risk cases does not expose the true picture of the various elements that contribute to surgical outcome and may be detrimental to training of future generations of cardiac surgeons. 1.Bridgewater B, Grayson A D, Jackson M et al. Surgeon specific mortality in adult cardiac surgery: comparison between crude and risk stratified data BMJ 2003;327:13-17. Competing interests: None declared |
|||
|
|
|||
|
Emanuela Taioli, Director of Research Milano 20122, Alessandro Ghirardini, Donato Greco, and Alessandro Nanni Costa
Send response to journal:
|
Quality improvement represents one the strategic targets for health care system in Italy, as indicated by the National Heath Plan (1); in this context, outcomes’ evaluation plays a major role. For this reason we read with interest the paper by Bridgewater et al. (2), as well as other papers and comments recently published on BMJ. In Italy we are currently conducting a national outcomes-report card project that was launched at the end of 2001 by the Ministry of Health (MOH) in conjunction with the National Institute of Health (ISS), regional health authorities, and various professional associations. The project is focused on outcomes after organ transplantations, hip replacement, radiotherapy after breast cancer and coronary artery by-pass. Outcome data for transplantations have already been analysed and available on the MOH web site (3), where an evaluation of the case mix effect was also reported for each transplant centre. Statistical analysis on coronary artery by-pass is currently being completed and includes data on 17,000 procedures performed in 60 public hospitals throughout Italy (4). Follow-up at thirty days for vital status is recorded directly by the participating centres: patients lost to follow -up are traced by contacting local vital register offices at the patients' place of residence and by reviewing death certificates. Mortality rates are calculated for each centre and compared by direct and indirect statistical adjustment: data will be available by the end of 2003 on the dedicated web site (www/bpac/iss.it). Our policy is to publish hospital-based report cards, rather than single surgeon-based mortality rates, as suggested by other Authors (5). However, we totally agree with Bridgewater that the complexity of cases is the strongest factor affecting mortality rates and, for this reason, we only present risk-stratified mortality rates. We believe that continuing evaluation and monitoring of quality data will be useful both for the improvement of hospital performance, and for developing a comprehensive information strategy for the patients, ensuring to consumers a reliable instrument for an informed choice. 1.http://www.ministerosalute.it/resources/static/psn/documenti/psn0305_2-3.pdf 2.Ben Bridgewater, Anthony D Grayson, Mark Jackson, Nicholas Brooks, Geir J Grotte, Daniel J M Keenan, Russell Millner, Brian M Fabri, and Mark Jones. Surgeon specific mortality in adult cardiac surgery: comparison between crude and risk stratified data. BMJ 2003; 327: 13-17. 3.http://www.ministerosalute.it/trapianti/qualita/dati_en.jsp 4.Seccareccia F., Capriani P., Diemoz S., Taioli E., Tosti ME., Greco D. Gruppo di Ricerca italiano "Progetto BPAC". Cross-sectional study of cardiac surgery centers within the "CABG Project" (short-term outcome in patients undergoing coronary artery bypass graft surgery in Italian cardiac surgery centers. Ital Heart J. 2003 Jan;4(1 Suppl):32-8. 5.JS Gilfillan. Ranking heart surgeons has pitfalls. BMJ, 2003,327:107 (12 July) Competing interests: None declared |
|||