Copyright by ICR Publishers 2005

Transcription

1 Does EuroSCORE Predict Length of Stay and Specific Postoperative Complications after Heart Valve Surgery? Ioannis K. Toumpoulis 1,2, Constantine E. Anagnostopoulos 1,2 1 Columbia University College of Physicians and Surgeons, Department of Cardiothoracic Surgery, St. Luke s - Roosevelt Hospital Center, New York, NY, USA, 2 Department of Cardiac Surgery, University of Athens School of Medicine, Attikon Hospital Center, Athens, Greece Background and aim of the study: EuroSCORE is the most rigorously evaluated scoring system in cardiac surgery. The study aim was to evaluate the performance of EuroSCORE in the prediction of in-hospital postoperative length of stay and specific major postoperative complications after heart valve surgery. Methods: Data obtained from 1,105 consecutive patients who underwent isolated or combined heart valve surgery were collected prospectively. The EuroSCORE model (standard and logistic) was used to predict in-hospital mortality, prolonged length of stay (>20 days) and major postoperative complications. A C statistic (receiver operating characteristic curve) was used to test discrimination of the EuroSCORE. Calibration of the model was assessed by the Hosmer-Lemeshow goodness-of-fit statistic. Results: In-hospital mortality was 6.3%, and 21.7% of patients had one or more majorcomplication. EuroSCORE showed very good discriminatory ability in predicting postoperative renal failure (C statistic: 0.78) and good discriminatory ability in predicting in-hospital mortality (C statistic: 0.72), prolonged length of stay (C statistic: 0.71), stroke over 24 h (C statistic: 0.73), gastrointestinal complications (C statistic: 0.73) and respiratory failure (C statistic: 0.71). There were no differences in terms of the discriminatory ability between standard and logistic EuroSCORE. The standard EuroSCORE model showed good calibration in predicting these outcomes (Hosmer-Lemeshow: p >0.05). The logistic EuroSCORE model showed good calibration, except for prolonged length of stay and respiratory failure. Conclusion: EuroSCORE can be used to predict not only in-hospital mortality, for which it was originally designed, but also prolonged length of stay and specific postoperative complications such as renal failure, stroke over 24 h, gastrointestinal complications and respiratory failure within the whole context of heart valve surgery. These outcomes can be predicted accurately using the standard EuroSCORE, which is very easily calculated. The Journal of Heart Valve Disease 2005;14: Risk stratification has become an essential element in the practice of cardiac surgery. Great progress has been made in identifying risk factors in order to adjust outcomes after coronary artery bypass grafting (CABG). Risk prediction for heart valve surgery has lagged behind that for CABG because there are fewer valve surgery cases than CABG cases, and therefore much more time has been taken to accumulate adequate numbers of patients from whom outcomes may be evaluated. Recently, however, several studies which included large numbers of patients undergoing heart valve surgery have provided a detailed analysis of risk Address for correspondence: Ioannis K. Toumpoulis MD, St. Luke s - Roosevelt Hospital Center at Columbia University, 515 West 59th Street, New York, NY, USA toumpoul@otenet.gr factors associated with heart valve operations to predict operative mortality (1-4). The European System for Cardiac Operative Risk Evaluation (EuroSCORE) is based on a large patient database drawn across Europe, and has been developed for the prediction of in-hospital mortality after cardiac surgery in adults (5). The EuroSCORE model has also been shown to provide good discrimination and calibration in predicting early mortality following heart valve surgery (6). The standard EuroSCORE was first introduced in 1999 (7), and since its validation in the Society of Thoracic Surgeons database (8) it has been increasingly adopted worldwide, mainly because of its ease of calculation. The full logistic model became available only recently, but might be a better predictor of risk, especially in high-risk patients (9). The aim of the present study was to evaluate and compare the performance of the standard and logistic Copyright by ICR Publishers 2005

2 244 EuroSCORE and heart valve surgery EuroSCORE models in predicting in-hospital mortality, prolonged postoperative length of stay and major postoperative complications in a series of 1,105 consecutive patients who underwent heart valve surgery at a single institution. In this study, major postoperative complications were evaluated both as one variable as well as separate variables, in order to test the performance of EuroSCORE in predicting specific complications. Clinical material and methods Patient population The study population comprised 1,105 consecutive adult patients who underwent isolated or combined valve surgery between January 1992 and March 2002 at the St. Luke s - Roosevelt Hospital Center affiliated with Columbia University. Registry databases were studied for preoperative, intraoperative and postoperative data of the patients. J Heart Valve Dis Data acquisition and definitions Data were collected prospectively during the patients admission as part of routine clinical practice, and entered into the New York State adult cardiac surgery report. Risk stratification was performed according to the standard and logistic EuroSCORE models. The factors used by the EuroSCORE formula, their definition and their score for both standard and logistic EuroSCORE are listed in Table I. Except for the variables utilized by the EuroSCORE model, postoperative data were also collected and included 30-day mortality, in-hospital mortality, postoperative length of stay and major complications after surgery: stroke, transmural myocardial infarction, deep sternal wound infection, re-exploration for bleeding, sepsis and/or endocarditis, gastrointestinal complications, renal failure and respiratory failure. A prolonged postoperative in-hospital length of stay (>20 days) was defined as exceeding approximately the 85th centile of its distribution. Stroke was defined as intraoperative stroke when a permanent new focal neurological deficit occurred intraoperatively to 24 h postoperatively, whilst the neurological deficit was defined as stroke over 24 h when it occurred more than 24 h postoperatively. Transmural myocardial infarction was defined by the appearance of new Q-waves on the electrocardiogram, combined with a rise in the creatine kinase MB isoenzyme (>100 µg/l). A diagnosis of deep sternal wound infection required at least one of the following criteria: (i) an organism was isolated from culture of mediastinal tissue or fluid; (ii) evidence of mediastinitis was seen during surgery; or (iii) one of the following: chest pain, sternal instability or fever (>38 C), was present and there was either purulent discharge from the mediastinum or an organism was isolated from blood culture or culture of drainage of the mediastinal area. Return to the operating room within 36 h postoperatively for reoperation to control bleeding or to evacuate large hematomas in the thorax or pericardium was defined as re-exploration for bleeding. Sepsis and/or endocarditis was defined as fever (>38 C), and two or more positive blood cultures related to the procedure associated with systemic signs of infection but without intracardiac localization (sepsis) and two or more positive blood cultures with demonstrated valvular vegetation, or acute valvular dysfunction caused by infection (endocarditis). These two variables were considered as a single postoperative complication in the New York State adult cardiac surgery report. Gastrointestinal complications considered any postoperative episode, whilst still in hospital, of vomiting blood, gross blood in the stool, or perforation or necrosis of the stomach or intestine. Postoperative renal failure was diagnosed in patients with normal renal function preoperatively who presented with a creatinine level >220 µmol/l for more than seven postoperative days, or who required temporary or permanent renal dialysis of any type. Finally, respiratory failure was diagnosed as pulmonary insufficiency requiring intubation and ventilation for a period of 72 h or more, at any time during the postoperative stay, whilst for patients who were placed on and taken off ventilation several times, the total of these episodes should be 72 h or more. Ethical issues The need for informed consent was waived, since the data used in this study had already been collected for clinical purposes. Furthermore, the present study did not interfere with the treatment of patients and the database was organized in a way that made identification of an individual patient impossible. Statistical methods Numerical variables were presented as mean ± SD, and discrete variables were summarized by percentages. A C statistic (receiver operating characteristic (ROC) curve) was used to assess the discriminatory ability of the standard and logistic EuroSCORE models. The area under the ROC curve (10) was calculated as an index for how well the EuroSCORE could discriminate between patients who lived and those who died, or between patients with a complication and those without this complication after cardiac surgery. The discriminative power of the model was considered excellent if the area under the ROC curve was >0.80, very good if >0.75, and good if >0.70 (11). Calibration of the model was assessed using the Hosmer- Lemeshow goodness-of-fit statistic (12). For this statis-

3 J Heart Valve Dis EuroSCORE and heart valve surgery 245 tic the predicted risks of individual patients were rankordered and divided into 10 groups of approximately equal size, based on their predicted probability. Within each group of estimated risk, the number of predicted deaths (or complications) was accumulated against the number of observed deaths (or complications); a p- value >0.05 indicated acceptable calibration of the model. All analyses were performed using SPSS 11.0 (SPSS, Inc., Chicago, IL, USA), and the p-values were two-tailed. Table I: Risk factors, definitions and weights in the standard (score) and logistic (β coefficient) EuroSCORE models. Risk factors Definition Score β coefficient Patient-related factors Age Per 5 years or part thereof over 60 years for standard EuroSCORE and continuous for logistic EuroSCORE Gender Female Chronic pulmonary disease Long-term use of bronchodilators or steroids for lung disease Extracardiac arteriopathy Any one or more of the following: claudication, carotid occlusion or >50% stenosis, previous or planned intervention on the abdominal aorta, limb arteries or carotids Neurological dysfunction Disease severely affecting ambulation or day-to-day functioning Previous cardiac surgery Requiring opening of the pericardium Serum creatinine >200 µmol/l preoperatively Active endocarditis Patient still under antibiotic treatment for endocarditis at the time of surgery Critical preoperative state Any one of the following: ventricular tachycardia or fibrillation or aborted sudden death, preoperative cardiac massage, preoperative ventilation before arrival in the anesthetic room, preoperative inotropic support, intraaortic balloon pump, or preoperative acute renal failure (anuria or oliguria <10 ml/h) Cardiac-related factors Unstable angina Rest angina requiring intravenous nitrates until arrival in the anesthetic room Left ventricular dysfunction Moderate or left ventricular ejection fraction 30-50% Left ventricular dysfunction Poor or left ventricular ejection fraction <30% Recent MI <90 days Pulmonary hypertension Systolic pulmonary pressure >60 mmhg Operation-related factors Emergent operation Carried out on referral before the beginning of the next working day Other than isolated CABG Major cardiac procedure other than or in addition to CABG Surgery on thoracic aorta For disorder of ascending, arch or descending aorta Post-infarct septal rupture Post-infarct septal rupture Constant β0 Only for logistic EuroSCORE CABG: Coronary artery bypass grafting; MI: Myocardial infarction.

4 246 EuroSCORE and heart valve surgery Results Table II: Heart valve surgical procedures. Heart valve procedure No. of patients Aortic valve surgery 285 (25.8) Mitral valve surgery 222 (20.1) Tricuspid valve surgery 5 (0.5) Double valve surgery 84 (7.6) Triple valve surgery 9 (0.8) CABG + Aortic valve surgery 282 (25.5) CABG + Mitral valve surgery 181 (16.4) CABG + Double valve surgery 37 (3.3) Values in parentheses are percentages. CABG: Coronary artery bypass grafting. Details of heart valve surgical procedures are listed in Table II. The median standard EuroSCORE was 8 (interquartile range, 6-11) and the median logistic EuroSCORE was (interquartile range, ). J Heart Valve Dis The mean age within the study sample was 66.1 ± 13.2 years. Patient and disease characteristics after allocating patients to six subgroups according to the factors utilized by standard EuroSCORE are listed in Table III. There was an increase in mean age and a percentage increase in risk factors and in patients with a lower ejection fraction as the risk stratification grew. Overall in-hospital mortality was 6.3% (n = 70), whilst in or out of hospital 30-day mortality was 5.7% (n = 63). The mean postoperative in-hospital length of stay was 15.2 ± 19.2 days, and 21.7% of the patients (n = 240) had at least one major complication. Details of early outcome among the six patient subgroups according to standard EuroSCORE are summarized in Table IV. There was an increase in in-hospital mortality, postoperative length of stay, prolonged length of stay and major complications as the risk stratification grew. Areas under the ROC curves both for standard and logistic EuroSCORE models for all outcomes analyzed are detailed in Table V. There were no differences between the standard and logistic EuroSCORE mod- Table III: Patient and disease characteristics. Patients were allocated to six subgroups according to their standard EuroSCORE. Variable EuroSCORE >14 n = 17 n = 231 n = 324 n = 295 n = 159 n = 79 Patient-related factors Age (years) * 50.5 ± ± ± ± ± ± 9.3 Female (n) 0 (0) 80 (34.6) 153 (47.2) 149 (50.5) 85 (53.5) 39 (49.4) Chronic pulmonary disease (n) 0 (0) 11 (4.8) 32 (9.9) 41 (13.9) 44 (27.7) 19 (24.1) Extracardiac arteriopathy (n) 0 (0) 1 (0.4) 12 (3.7) 30 (10.2) 42 (26.4) 40 (50.6) Neurological dysfunction (n) 0 (0) 6 (2.6) 19 (5.9) 28 (9.5) 16 (10.1) 21 (26.6) Previous cardiac surgery (n) 0 (0) 5 (2.2) 43 (13.3) 61 (20.7) 38 (23.9) 28 (35.4) Serum creatinine >200 µmol/l (n) 0 (0) 2 (0.9) 7 (2.2) 22 (7.5) 18 (11.3) 25 (31.6) Active endocarditis (n) 0 (0) 6 (2.6) 15 (4.6) 13 (4.4) 4 (2.5) 8 (10.1) Critical preoperative state (n) 0 (0) 0 (0) 4 (1.2) 16 (5.4) 27 (17.0) 31 (39.2) Cardiac-related factors Unstable angina (n) 0 (0) 20 (8.7) 91 (28.1) 167 (56.6) 123 (77.4) 68 (86.1) Ejection fraction >50% (n) 17 (100) 100 (43.3) 103 (31.8) 55 (18.6) 13 (8.2) 6 (7.6) Ejection fraction 30-50% (n) 0 (0) 123 (53.2) 195 (60.2) 182 (61.7) 102 (64.1) 43 (54.4) Ejection fraction <30% (n) 0 (0) 8 (3.5) 26 (8.0) 58 (19.7) 44 (27.7) 30 (38.0) Recent MI (n) 0 (0) 2 (0.9) 14 (4.3) 41 (13.9) 64 (40.3) 45 (57.0) Pulmonary hypertension (n) 0 (0) 4 (1.7) 28 (8.6) 31 (10.5) 37 (23.3) 30 (38.0) Operation-related factors Emergency (n) 0 (0) 1 (0.4) 4 (1.2) 12 (4.1) 10 (6.3) 26 (32.9) Other than isolated CABG (n) 17 (100) 231 (100) 324 (100) 295 (100) 159 (100) 79 (100) Surgery on thoracic aorta (n) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Post-infarct septal rupture (n) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) * Values are mean ± SD. Values in parentheses are percentages. CABG: Coronary artery bypass grafting; MI: Myocardial infarction.

5 J Heart Valve Dis EuroSCORE and heart valve surgery 247 Table IV: Early mortality, postoperative length of stay and major complications in the six subgroups of standard EuroSCORE. Variable EuroSCORE >14 Total n = 17 n = 231 n = 324 n = 295 n = 159 n = 79 n = 1,105 Early mortality and length of stay 30-day mortality (n) 0 (0) 4 (1.7) 14 (4.3) 19 (6.4) 11 (6.9) 15 (19.0) 63 (5.7) In-hospital mortality (n) 0 (0) 5 (2.2) 12 (3.7) 20 (6.8) 18 (11.3) 15 (19.0) 70 (6.3) Postop. length of stay (days) * 8.1 ± ± ± ± ± ± ± 19.2 Postop. length of stay >20 days (n) 0 (0) 14 (6.1) 36 (11.1) 61 (20.7) 48 (30.2) 30 (38.0) 189 (17.1) Major complications Patients with complications (n) 1 (5.9) 25 (10.8) 51 (15.7) 77 (26.1) 53 (33.3) 33 (41.8) 240 (21.7) Intraoperative stroke (n) 0 (0) 3 (1.3) 9 (2.8) 12 (4.1) 10 (6.3) 7 (8.9) 41 (3.7) Over 24 h stroke (n) 0 (0) 1 (0.4) 1 (0.3) 3 (1.0) 3 (1.9) 4 (5.1) 12 (1.1) Postoperative MI (n) 1 (5.9) 1 (0.4) 1 (0.3) 0 (0) 0 (0) 1 (1.3) 4 (0.4) Deep sternal wound infection (n) 0 (0) 1 (0.4) 0 (0) 2 (0.7) 0 (0) 1 (1.3) 4 (0.4) Bleeding/reoperation (n) 0 (0) 8 (3.5) 9 (2.8) 24 (8.1) 11 (6.9) 3 (3.8) 55 (5.0) Sepsis/endocarditis (n) 0 (0) 1 (0.4) 9 (2.8) 5 (1.7) 8 (5.0) 7 (8.9) 30 (2.7) Gastrointestinal complications (n) 0 (0) 0 (0) 3 (0.9) 4 (1.4) 3 (1.9) 4 (5.1) 14 (1.3) Renal failure (n) 0 (0) 0 (0) 2 (0.6) 9 (3.1) 2 (1.3) 8 (10.1) 21 (1.9) Respiratory failure (n) 0 (0) 6 (2.6) 16 (4.9) 33 (11.2) 26 (16.4) 11 (13.9) 92 (8.3) * Values are mean ± SD. Values in parentheses are percentages. MI: Myocardial infarction. els. EuroSCORE showed very good discriminatory ability in predicting postoperative renal failure, in addition to a good discriminatory ability to predict inhospital mortality (Fig. 1), prolonged postoperative length of stay, stroke over 24 h, gastrointestinal complications and respiratory failure. The Hosmer- Lemeshow tests of calibration for standard and logistic EuroSCORE models are listed in Table VI. All p-values in the standard EuroSCORE model were not statistically significant (p >0.05), indicating good calibration in predicting in-hospital mortality, prolonged length of stay (>20 days), stroke over 24 h, gastrointestinal complications, renal failure and respiratory failure. The logistic EuroSCORE showed good calibration in predicting these outcomes, except for prolonged length of stay and respiratory failure. Table V: C statistics (ROC area under the curve; 95% confidence intervals in parentheses) of the standard and logistic EuroSCORE models. Variable Standard EuroSCORE Logistic EuroSCORE In-hospital mortality 0.72 ( ) 0.72 ( ) Postop. length of stay >20 days 0.71 ( ) 0.71 ( ) Postop. complications 0.66 ( ) 0.66 ( ) Intraoperative stroke 0.67 ( ) 0.67 ( ) Stroke over 24 h 0.73 ( ) 0.73 ( ) Postop. MI 0.38 ( ) 0.41 ( ) Deep sternal wound infection 0.59 ( ) 0.56 ( ) Bleeding requiring reoperation 0.58 ( ) 0.58 ( ) Sepsis and/or endocarditis 0.68 ( ) 0.68 ( ) Gastrointestinal complications 0.73 ( ) 0.73 ( ) Renal failure 0.78 ( ) 0.77 ( ) Respiratory failure 0.71 ( ) 0.71 ( ) MI: Myocardial infarction.

6 248 EuroSCORE and heart valve surgery J Heart Valve Dis Table VI: Calibration (goodness-of-fit, Hosmer-Lemeshow) of standard and logistic EuroSCORE model of the predicted outcomes with good discriminatory ability. Variable Standard EuroSCORE Logistic EuroSCORE In-hospital mortality χ 2 (8 df) = 7.41, p = χ 2 (8 df) = 10.30, p = Postop. length of stay >20 days χ 2 (8 df) = 10.58, p = χ 2 (8 df) = 25.07, p = Stroke over 24 h χ 2 (8 df) = 10.51, p = χ 2 (8 df) = 4.79, p = Gastrointestinal complications χ 2 (8 df) = 5.19, p = χ 2 (8 df) = 6.80, p = Renal failure χ 2 (8 df) = 8.16, p = χ 2 (8 df) = 7.05, p = Respiratory failure χ 2 (8 df) = 9.23, p = χ 2 (8 df) = 31.19, p <0.001 Discussion Figure 1: Receiver operating characteristic (ROC) curves for in-hospital mortality. There was no difference in the area under the ROC curve between the standard and logistic EuroSCORE models. Most predictive models in cardiac surgery use early mortality as an end-point. Only a few models evaluate other relevant outcomes such as morbidity and postoperative in-hospital length of stay or length of stay in the intensive care unit (13-16). The EuroSCORE, however, was developed to score mortality during hospital stay (5). In addition, although large studies have focused on risk stratification for heart valve surgery, they only evaluated early mortality as an end-point (1-4). A number of studies have been conducted that have tested the accuracy of EuroSCORE in predicting postoperative morbidity after cardiac surgery (13,16,17), though with contradictory results depending on the parameters selected. However, risk stratification and preoperative prediction of specific postoperative complications and prolonged length of stay appears to be desirable, especially when this can be performed by already widely used risk stratification models. Despite substantial demographic differences between Europe and North America, EuroSCORE performs very well in the Society of Thoracic Surgeons database (8), and this was confirmed in the present study, where the discriminatory ability of standard and logistic EuroSCORE models in predicting in-hospital mortality (C statistic: 0.72) was good, whilst in the original EuroSCORE database the C statistic was 0.75 for the prediction of early mortality after heart valve surgery (6). It was also demonstrated that the standard and logistic EuroSCORE models were each well calibrated and showed no difference in their discriminatory abilities (Fig. 1). The EuroSCORE model is based on 17 preoperative risk factors, and does not take into consideration any possible negative intraoperative events such as prolonged cross-clamp time, cardiopulmonary bypass time and requirement for mechanical support on completion of the procedure, all of which have been proved to be strong predictors for postoperative morbidity after cardiac surgery (17). However, the EuroSCORE model performs very well generally in cardiac surgery, or specifically in CABG or in heart valve surgery. This performance may be explained by the fact that, overall, the risk factors for short-term mortality following heart valve surgery or CABG appear to be relatively consistent (2). In addition, it was shown in the present study that there was a similar correlation in risk factors affecting in-hospital mortality, prolonged length of stay and specific postoperative complications after heart valve surgery. This suggested the possibility of using a single risk stratification system to predict multiple outcomes within the whole context of heart valve surgery. Overall, EuroSCORE is the best-established and validated risk model for contemporary practice in cardiac surgery (18). Recently, the results of one study showed that EuroSCORE could be correlated to the costs of cardiac surgery (19), whilst the present authors showed that both standard and logistic EuroSCORE models could be used to predict long-term mortality after

7 J Heart Valve Dis EuroSCORE and heart valve surgery 249 CABG (20). The identification of high-risk patients for long-term mortality is very important in order to ensure that these patients will have more frequent follow up examinations, as well as appropriate conservative therapy. Similarly, Kasimir et al. (21) reported that EuroSCORE could predict mid-term outcome after combined valve and CABG. In the present study, it was shown clearly that EuroSCORE could be used to predict prolonged postoperative length of stay and postoperative renal failure, stroke over 24 h, gastrointestinal complications and respiratory failure within the whole context of heart valve surgery. The preventive process for many postoperative complications starts with the stratification of patients into high- and low-risk groups. Thus, the identification of patients at high risk of developing the above devastating complications might help to reduce their numbers by bringing the patient preoperatively to an optimal condition (renal function, chronic obstructive pulmonary disease, etc.) and by modifying the surgical procedure to achieve the greatest benefit. The ability of a single model such as the standard EuroSCORE - which can be calculated at the bedside - to predict accurately all of the above-mentioned outcomes renders it a powerful tool in everyday clinical heart valve surgical practice, both for the surgeon and the patient. Study limitations The first limitation was that despite the present study being retrospective in design, the data were collected prospectively, using highly standardized methods of the New York State audited database. Second, the study referred to a single-center regional database with a relatively small number of patients; thus, the results require further evaluation before being generalized across diverse institutions and countries. The authors intention was neither to develop a new score nor to investigate the impact of individual variables on prolonged length of stay and specific postoperative complications. It is known that, for most risk factors, the predictive value for mortality differs considerably from that for morbidity (13), whilst the inclusion of intraoperative parameters in a model could further increase its discriminatory ability (17). In conclusion, the standard and logistic EuroSCORE models can be used to predict postoperative renal failure with very good discriminatory ability, as well as inhospital mortality, prolonged postoperative length of stay (>20 days), stroke over 24 h, gastrointestinal complications and respiratory failure. The standard EuroSCORE model was well-calibrated in predicting all these outcomes, and this model may be useful for simple bed-side calculations within the whole context of heart valve surgery. References 1. Edwards FH, Peterson ED, Coombs LP, et al. Prediction of operative mortality after valve replacement surgery. J Am Coll Cardiol 2001;37: Gardner SC, Grunwald GK, Rumsfeld JS, et al. Comparison of short-term mortality risk factors for valve replacement versus coronary artery bypass graft surgery. Ann Thorac Surg 2004;77: Jamieson WR, Edwards FH, Schwartz M, Bero JW, Clark RE, Grover FL. Risk stratification for cardiac valve replacement. National Cardiac Surgery Database. Database Committee of The Society of Thoracic Surgeons. Ann Thorac Surg 1999;67: Nowicki ER, Birkmeyer NJ, Weintraub RW, et al. Multivariable prediction of in-hospital mortality associated with aortic and mitral valve surgery in Northern New England. Ann Thorac Surg 2004;77: Roques F, Nashef SA, Michel P, et al. Risk factors and outcome in European cardiac surgery: Analysis of the EuroSCORE multinational database of patients. Eur J Cardiothorac Surg 1999;15: Roques F, Nashef SA, Michel P. Risk factors for early mortality after valve surgery in Europe in the 1990s: Lessons from the EuroSCORE pilot program. J Heart Valve Dis 2001;10: Nashef SA, Roques F, Michel P, Gauducheau E, Lemeshow S, Salamon R. European system for cardiac operative risk evaluation (EuroSCORE). Eur J Cardiothorac Surg 1999;16: Nashef SA, Roques F, Hammill BG, et al. Validation of European System for Cardiac Operative Risk Evaluation (EuroSCORE) in North American cardiac surgery. Eur J Cardiothorac Surg 2002;22: Michel P, Roques F, Nashef SA. Logistic or additive EuroSCORE for high-risk patients? Eur J Cardiothorac Surg 2003;23: Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143: Swets JA. Measuring the accuracy of diagnostic systems. Science 1988;240: Hosmer DW, Taber S, Lemeshow S. The importance of assessing the fit of logistic regression models: A case study. Am J Public Health 1991;81: Geissler HJ, Holzl P, Marohl S, et al. Risk stratification in heart surgery: Comparison of six score systems. Eur J Cardiothorac Surg 2000;17: Immer F, Habicht J, Nessensohn K, et al. Prospective evaluation of 3 risk stratification scores in cardiac surgery. Thorac Cardiovasc Surg 2000;48:

8 250 EuroSCORE and heart valve surgery 15. Kurki TS, Hakkinen U, Lauharanta J, Ramo J, Leijala M. Evaluation of the relationship between preoperative risk scores, postoperative and total length of stays and hospital costs in coronary bypass surgery. Eur J Cardiothorac Surg 2001;20: Pitkanen O, Niskanen M, Rehnberg S, Hippelainen M, Hynynen M. Intra-institutional prediction of outcome after cardiac surgery: Comparison between a locally derived model and the EuroSCORE. Eur J Cardiothorac Surg 2000;18: Stoica SC, Sharples LD, Ahmed I, Roques F, Large SR, Nashef SA. Preoperative risk prediction and intraoperative events in cardiac surgery. Eur J Cardiothorac Surg 2002;21: Gogbashian A, Sedrakyan A, Treasure T. J Heart Valve Dis EuroSCORE: A systematic review of international performance. Eur J Cardiothorac Surg 2004;25: Pinna PP, Bobbio M, Colangelo S, Veglia F, Marras R, Diena M. Can EuroSCORE predict direct costs of cardiac surgery? Eur J Cardiothorac Surg 2003;23: Toumpoulis IK, Anagnostopoulos CE, DeRose JJ, Swistel DG. European system for cardiac operative risk evaluation predicts long-term survival in patients with coronary artery bypass grafting. Eur J Cardiothorac Surg 2004;25: Kasimir MT, Bialy J, Moidl R, et al. EuroSCORE predicts mid-term outcome after combined valve and coronary bypass surgery. J Heart Valve Dis 2004;13: