Although some renal dysfunction is invariably associated

Transcription

1 Valvular Heart Operation Is an Independent Risk Factor for Acute Renal Failure Antony D. Grayson, Magdy Khater, FRCA, Mark Jackson, PhD, and Mark A. Fox, FRCA Departments of Research and Development and Cardiothoracic Anaesthesia, The Cardiothoracic Centre, Liverpool, United Kingdom Background. Acute renal failure (ARF) after cardiac operation with cardiopulmonary bypass is associated with a high mortality rate. The purpose of this study was to determine and quantify whether valvular heart operation is an independent risk factor for developing ARF. Methods. We retrospectively analyzed 5,132 consecutive patients who underwent cardiac operation involving cardiopulmonary bypass between April 1997 and March Patients with significant renal impairment (preoperative serum creatinine > 200 mol/l) were excluded. A multivariable logistic regression model was constructed to identify independent risk factors for the postoperative development of ARF. Results. In 151 (2.9%) patients ARF developed before hospital discharge. The crude incidence of ARF for isolated coronary artery bypass grafting, isolated valve(s) operation, and valve(s) with coronary artery bypass grafting operation was 1.9%, 4.4%, and 7.5%, respectively (p < 0.001). The results of the logistic regression analysis found that valve operation with or without coronary artery bypass grafting was an independent risk factor for the development of postoperative ARF (odds ratio 2.68, 95% confidence interval 1.89 to 3.79; p < 0.001). Other independent predictors of ARF were increased preoperative serum creatinine levels, urgent or emergent operation, insulin-dependent diabetes, and increased cardiopulmonary bypass time. Conclusions. Valve operation is an independent risk factor for postoperative ARF. This risk is further increased by prolonged cardiopulmonary bypass. (Ann Thorac Surg 2003;75: ) 2003 by The Society of Thoracic Surgeons Although some renal dysfunction is invariably associated with cardiac operation involving cardiopulmonary bypass [1], renal reserve is usually sufficient to prevent this from becoming clinically significant. However, acute renal failure (ARF) remains a serious complication after cardiac operation performed with cardiopulmonary bypass and carries a high mortality rate. Acute renal failure has been reported to range from 1% to 30%, depending on the definition used, with a mortality rate between 7% and 38% [2 7]. When ARF requires hemodialysis, the associated mortality rate can exceed 60% [4, 8]. A greater understanding of the etiology of this condition will permit more active preventive management. We have noted through clinical observation that patients who had valvular cardiac operation appear to be at greater risk of developing ARF. We therefore aimed to determine and quantify the independent effect of valvular operation on postoperative ARF. Material and Methods Patient Population and Data Data were collected prospectively on a total of 5,132 adult patients who had a cardiac operation between April 1, 1997 and March 31, 2001 at the Cardiothoracic Center- Accepted for publication Jan 16, Address reprint requests to Mr Grayson, Regional Clinical Information Analyst, The Cardiothoracic Centre-Liverpool, Thomas Drive, Liverpool L14 3PE, United Kingdom; tony.grayson@ctc.nhs.uk. Liverpool. Patients who had coronary artery bypass grafting (CABG) with or without heart valve repair or replacement that was incidental to resection of a ventricular aneurysm, or other surgical procedures were not included. We also excluded patients with a preoperative serum creatinine level greater than 200 mol/l, history of renal dysfunction, or who underwent operation without cardiopulmonary bypass. Data collection methods and definitions have been described in detail previously [9]. Data were collected during the patients admission as part of routine clinical practice. Data included patient characteristics (Table 1), preoperative medications (Table 2), and operative characteristics (Table 3). Outcome variables collected included ARF, in-hospital mortality rate, and length of postoperative hospital stay. Acute renal failure was defined postoperative serum creatinine level greater than 200 mol/l or patients requiring dialysis therapy or hemofiltration before hospital discharge. In-hospital death was defined as death within the same hospital admission regardless of cause. All patients transferred from the base hospital to another hospital were followed up to confirm their status at discharge. Statistical Methods Because of the nonnormality of continuous variables they are reported as median with 25th and 75th percentiles. Categorical variables are reported as a percentage with 95% confidence intervals (CI). Comparisons were made with Wilcoxon rank-sum test and 2 test as appropriate by The Society of Thoracic Surgeons /03/$30.00 Published by Elsevier Inc PII S (03)

2 1830 GRAYSON ET AL Ann Thorac Surg VALVULAR OPERATION AS RISK FACTOR FOR ARF 2003;75: Table 1. Univariate Association Between Patient Characteristics and Acute Renal Failure % of Patients ARF (%) Age (years) Reference ( ) ( ) Body mass index (kg/m 2 ) Reference ( ) Gender Male Reference Female ( ) Angina Class Asymptomatic/I Reference II ( ) III ( ) IV ( ) NYHA Class I Reference II ( ) III ( ) IV ( ) Diabetes No Reference Diet/oral ( ) Insulin-dependent ( ) Preoperative serum creatinine ( mol/l) Reference ( ) ( ) Hypertension No Reference Yes ( ) Prior myocardial infarction No Reference Yes ( ) Congestive cardiac failure No Reference Yes ( ) Current smoker No Reference Yes ( ) Respiratory disease No Reference Yes ( ) Peripheral vascular disease No Reference Yes ( ) Cerebro vascular disease No Reference Yes ( ) 0.551

3 Ann Thorac Surg GRAYSON ET AL 2003;75: VALVULAR OPERATION AS RISK FACTOR FOR ARF 1831 Table 1. Continued % of Patients ARF (%) Carotid bruit No Reference Yes ( ) Prior percutaneous coronary intervention No Reference Yes ( ) Prior cardiac operation No Reference Yes ( ) Ejection fraction Good ( 50%) Reference Moderate (30% 49%) ( ) Poor ( 30%) ( ) Extent of disease Normal Reference One vessel ( ) Two vessel ( ) Three vessel ( ) Left main stem stenosis 50% No Reference Yes ( ) Priority Elective Reference Urgent ( ) Emergent ( ) ARF acute renal failure; CI confidence interval; NYHA New York Heart Association. Standard statistical tests were used to calculate odds ratios (OR) and 95% CI. A multivariable logistic regression analysis was undertaken to identify independent risk factors for ARF [10, 11]. All variables in Tables 1, 2, and 3 were included as potential risk factors in the Table 2. Univariate Association Between Preoperative Medications and Acute Renal Failure %of Patients ARF (%) Digoxin No Ref. Yes ( ) Diuretic No Ref. Yes ( ) ACE inhibitors No Ref. Yes ( ) Intravenous nitrates No Ref. Yes ( ) ARF acute renal fail- ACE angiotensin-converting enzyme; ure; CI confidence interval. logistic regression model, including interactions between variables such as type of procedure and duration of cardiopulmonary bypass. The C statistic (equivalent to the area under the receiver operating characteristic curve) and the Hosmer-Lemeshow goodness-of-fit statistic were calculated to assess the performance and calibration of the model, respectively [11, 12]. In all cases a p value less than 0.05 was considered significant. All statistical analyses were performed retrospectively with SAS for Windows version 8 (SAS Institute, Cary, NC). Results Overall, of the 5,132 patients in the study, ARF developed in 151 patients (2.9% [95% CI 2.5% to 3.4%]) after cardiac operation and before discharge from the hospital. One hundred five (2.0% [95% CI 1.7% to 2.5%]) patients with ARF did not require dialysis, compared with 46 (0.9% [95% CI 0.7 to 1.2]) patients who had ARF that required dialysis. The patients classified as having ARF with a postoperative serum creatinine level greater than 200 mol/l without requiring dialysis had an average preoperative serum creatinine of 108 mol/l (minimum 73 mol/l and maximum 127 mol/l).

4 1832 GRAYSON ET AL Ann Thorac Surg VALVULAR OPERATION AS RISK FACTOR FOR ARF 2003;75: Table 3. Univariate Association Between Operative Characteristics and Acute Renal Failure %of Patients ARF (%) Cardiac procedure CABG only Reference Valve(s) only ( ) CABG and valve(s) ( ) Cardiopulmonary bypass duration (min) Reference ( ) ( ) Aortic cross clamp duration (min) Reference ( ) ( ) ARF acute renal failure; CABG coronary artery bypass grafting; CI confidence interval. Three thousand seven hundred forty-two patients (72.9% [95% CI 71.7% to 74.1%]) had isolated CABG, compared with 855 (16.7% [95% CI 15.6% to 17.7%]) with isolated valve procedures and 535 (10.4% [95% CI 9.6% to 11.3%]) with combined CABG and valve procedures. Table 1 shows patient characteristics and the association with ARF. In the univariate analysis, the preoperative characteristics that were significantly associated with the development of ARF included advanced age, greater severity of angina and dyspnea, insulin-dependent diabetes, higher serum creatinine level, congestive cardiac failure, respiratory disease, peripheral vascular disease, previous cardiac operation, decreased left ventricular ejection fraction, and priority of operation. Preoperative digoxin, diuretics, and angiotensin-converting enzyme inhibitors were also significantly associated with postoperative ARF (Table 2). In the univariate analysis, valve operation with or without CABG increased the odds of postoperative ARF (Table 3). The associated risk between valve operation and ARF increased with the number of valves (single valve 5.5%, double valve 7.6%), although this failed to reach statistical significance (p 0.50). No association could be found in relation to whether the type of valve operation (ie, repair or replacement valve operation) influenced the development of ARF. However, there was a trend to suggest that a single mitral valve operation had a higher incidence of ARF than single aortic valve operation (7.6% versus 5%, p 0.13). Patients with prolonged cardiopulmonary bypass and aortic cross-clamp times also had increased odds of ARF (Table 3). Cardiopulmonary bypass times were significantly longer (p 0.001) in combined CABG and valve procedures (median 146 minutes [25th and 75th percentiles 122 to 171 minutes]) than in isolated valve operation (median 103 minutes [25th and 75th percentiles 80 to 121 minutes]) or isolated CABG (median 105 minutes [25th and 75th percentiles 87 to 126 minutes]). Overall, valve operation with or without CABG on average lasted 114 minutes (25th and 75th percentiles 89 to 147 minutes) Table 4. Independent Risk Factors for Acute Renal Failure Coefficient SE Adjusted (95% CI) P Value Preoperative serum creatinine ( mol/l) a ( ) Valvular Procedure CABG ( ) Urgent or emergent procedure ( ) Insulin-dependent diabetes ( ) CPB duration (hours) b ( ) Intercept a For every additional 10 mol/l. b For every additional hour. Calculation of predicted risk using patient data and logistic regression coefficients: Calculate the odds of acute renal failure exp{ [ preoperative serum creatinine (continuous value)] [ valvular operation] [ nonelective operation] [ insulin-dependent diabetes] [ CPB duration (continuous value)]}. Predicted risk of acute renal failure as a percentage [odds/(1 odds)] 100. CABG coronary artery bypass grafting; CI confidence interval; CPB cardiopulmonary bypass; SE standard error.

5 Ann Thorac Surg GRAYSON ET AL 2003;75: VALVULAR OPERATION AS RISK FACTOR FOR ARF 1833 Fig 1. Risk of acute renal failure (ARF) stratified by procedure as preoperative serum creatinine increases. The solid lines represent the predicted risk of acute renal failure and the dashed lines represent the associated 95% confidence intervals. (CABG coronary artery bypass grafting.) compared with 105 minutes (25th and 75th percentiles 87 to 126 minutes) for isolated CABG (p 0.001). The results of the logistic regression analysis found that valve operation with or without CABG was an independent risk factor for the development of postoperative ARF (OR 2.68, 95% CI 1.89 to 3.79; p 0.001). Other independent predictors of ARF were higher preoperative creatinine level, urgent or emergent operation, insulin-dependent diabetes, and longer cardiopulmonary bypass time. These variables are summarized in Table 4, with their regression coefficient, adjusted odds ratios, and p values. The discriminatory ability of the logistic model, as measured by the C statistic, was 0.78, indicating a good ability to discriminate between patients who developed ARF and those who did not. The predicted risks of individual patients were rank-ordered and divided into deciles. Within each decile of estimated risk, the number of ARFs predicted was compared with the number of observed ARFs. The Hosmer-Lemeshow goodness-of-fit statistic across deciles of risk was not statistically significant (p 0.129), indicating little departure from a perfect fit. Table 4 shows an example of the calculation of predicted risk for an individual patient. Using the logistic regression equation, Figures 1 and 2 show the increased risk of ARF, with associated 95% confidence intervals, stratified by procedure as preoperative serum creatinine and cardiopulmonary bypass time increase. The incidence of in-hospital death after isolated CABG for patients with ARF was 32.9% (95% CI 22.6% to 44.9%), compared with 1.7% (95% CI 1.3% to 2.1%) for patients without ARF (p 0.001). The median postoperative length of stay in survivors after isolated CABG for patients with ARF was 17 days (25th and 75th percentiles 10 to 34 days) compared with 8 days (25th and 75th percentiles 7 to 9 days) for patients without ARF (p 0.001). The incidence of in-hospital death after valve operation with or without CABG for patients with ARF was 46.2% (95% CI 34.9% to 57.7%) compared with 3.4% (95% CI 2.5% to 4.6%) for patients without ARF (p 0.001). The median postoperative length of stay in survivors after valve operation with or without CABG for patents with ARF was 20 days (25th and 75th percentiles 15 to 44 days) compared with 9 days (25th and 75th percentiles 7 to 11 days) for patients without ARF (p 0.001). Comment Acute renal failure after cardiac operation, necessitating dialysis, has a profound effect on survival. Chertow and colleagues [4] reported a mortality rate of 63.7% in patients with ARF requiring dialysis compared with a mortality rate of 4.3% in those without ARF. Several investigators have attempted to determine which factors predict development of ARF, with the expectation that by modifying the factors identified the incidence of ARF will decrease. Several studies have lacked a large enough sample size to reach firm conclusions [6, 13 15]. Notable exceptions are the studies by Chertow and colleagues [4], Conlon and associates [2], Mangano and colleagues [3], Andersson and associates [5], and Frost and coworkers [8]. In the multivariable logistic regression analysis we confirmed that valvular heart operation with or without CABG confers a 2.68-fold higher risk of ARF than isolated CABG, independent of other risk factors. This finding was adjusted for preoperative serum creatinine, nonelective operation, insulin-dependent diabetes, and most importantly independent of duration of cardioplumonary bypass. The finding that valve operation is a risk factor for ARF is not surprising, as we would expect that most complications are higher in valve operation compared with isolated CABG. This study, however, quantifies by how much more the risk of ARF is increased, with a prediction equation which may be useful to clinicians in assessing the risk of a patient developing this severe complication. Chertow and colleagues, in a landmark study involving 42,773 predominately white men who had cardiac oper- Fig 2. Risk of acute renal failure (ARF) stratified by procedure as cardiopulmonary bypass time increases. The solid lines represent the predicted risk of acute renal failure and the dashed lines represent the associated 95% confidence intervals. (CABG coronary artery bypass grafting.)

6 1834 GRAYSON ET AL Ann Thorac Surg VALVULAR OPERATION AS RISK FACTOR FOR ARF 2003;75: ation at 43 Department of Veterans Affairs medical centers, also showed that valve operation was an independent predictor for ARF. Other independent risk factors for the development of ARF included preoperative renal dysfunction, intraaortic balloon pump support, previous heart operation, preoperative heart failure, peripheral vascular disease, and chronic obstructive airway disease. Their study had relatively few women, and the authors also acknowledged that extremes of age and patients with severe diabetes mellitus were both underrepresented in the cohort studied [4]. As with other reports [2, 3, 5, 8], higher preoperative serum creatinine level was an independent risk factor for ARF in our study. Frost and colleagues [8], in a study involving 1,988 patients, found that preoperative serum creatinine greater than 110 mol/l significantly increased the risk of developing postoperative ARF. In our own experience, preoperative serum creatinine more than 100 mol/l significantly increased the risk of ARF (Fig 1). Our experience agrees with the work done at the Duke University Medical Center in Durham North Carolina, in which diabetes was identified as a risk factor for ARF [2, 16]. This finding is also supported by Mangano and colleagues [3]. However, these studies did not look at which type of diabetes influenced ARF. We identified insulin-dependent diabetes as an independent predictor, but not diet or oral-medication controlled diabetes. As with other reports [2, 3, 17, 18], our study identified prolonged cardiopulmonary bypass duration as a significant risk factor for postoperative ARF. Conlon and associates [2], in their study involving 2,848 patients who underwent cardiac operation, observed a linear relationship between duration on bypass and ARF. They concluded that shortening the time on bypass might help reduce the risk of ARF, but they did not suggest how this might be achieved, nor did they propose a mechanism to explain the phenomenon. Because of the susceptibility of the brain to clinically evident embolic damage, much attention has been focused on this organ. The kidney is also susceptible to embolic damage. Deal and colleagues [19] have shown that embolic load to an organ is proportional to the cardiac output that organ receives. Subsequently they [20] showed that cardiotomy suction during cardiopulmonary bypass can also produce embolic injury in a dog model. It seems likely that the increased cardiotomy suction and other embolic loads that are known to be associated with valve operation could, at least in part, explain the independent predictive power of valve operation for the development of ARF. This mechanism would also provide an explanation for the predictive effect of prolonged cardiopulmonary bypass (with its associated increased cardiotomy suction). Although our sample size was relatively large, there are some limitations that might affect the findings drawn from our observational study of patients who underwent cardiac operation. Using a threshold definition of more than 200 mol/l of postoperative serum creatinine for ARF may imply that some patients had a trivial baseline increase in serum creatinine resulting in a classification of ARF. However, all our patients classified as ARF, without requiring dialysis, had a baseline serum creatinine increase of 73 mol/l. This is comparable with the findings of Mangano and colleagues [3], who regarded anyone with a serum creatinine increase of 62 mol/l or more over baseline as having clinically significant ARF. Increased age was a univariate risk factor for ARF; however, our study did not show increased age as an independent predictor for ARF after cardiopulmonary bypass, as other studies have suggested [2, 3, 5, 16]. This could result from possible selection bias, with older patients not being offered cardiac operation. Aprotinin containing preservative also contributes to ARF, whereas preservative-free aprotinin is not associated with ARF [21, 22]. The use of aprotinin in this series was less than 1% and was entirely preservative free. Multivariable analysis is not a substitute for a welldesigned randomized control trial. The retrospective nature of the study cannot account for the unknown variables affecting the outcome that are not measured in this study. Conversely, retrospective comparisons with multivariable analysis are more versatile and more widely applicable than randomized control trials. In conclusion, we have shown that valve operation, with or without CABG, carries a higher risk of acute renal failure than isolated CABG. Other risk factors included higher preoperative serum creatinine level, urgent or emergent operation, insulin-dependent diabetes, and longer cardiopulmonary bypass time. Furthermore, we have quantified this risk with a prediction equation, which might prove useful in assessing the probability of a patient developing this severe complication. We would like to acknowledge the cooperation given to us by all the Consultant Cardiac Surgeons at the Cardiothoracic Centre- Liverpool: Mr John A. C. Chalmers, Mr Walid C. Dihmis, Mr M. John Drakeley, Mr Brian M. Fabri, Miss Elaine M. Griffiths, Mr Neeraj K. Mediratta, Mr Richard D. Page, Mr D. Mark Pullan, Mr Abbas Rashid, and Mr W. Ian Weir. We also would like to thank Janet Deane, who maintains the quality and ensures completeness of data collected in our Cardiac Surgery Registry. References 1. Ip-Yam PC, Murphy S, Baines M, Fox MA, Desmond MJ, Innes PA. Renal function and proteinuria after cardiopulmonary bypass: the effects of temperature and mannitol. Anesth Analg 1994;78: Conlon PJ, Stafford-Smith M, White WD, et al. Acute renal failure following cardiac surgery. Nephrol Dial Transplant 1999;14: Mangano CM, Diamondstone LS, Ramsay JG, Aggarwal A, Herskowitz A, Mangano DT. Renal dysfunction after myocardial revascularisation: risk factors, adverse outcomes, and hospital resource utilization. The Multicenter Study of Perioperative Ischemia Research Group. Ann Intern Med 1998; 128: Chertow GM, Lazarus JM, Christiansen CL, et al. Preoperative renal risk stratification. Circulation 1997;95: Andersson LG, Ekroth R, Bratteby LE, Hallhagen S, Wesslen O. Acute renal failure after coronary surgery a study of

7 Ann Thorac Surg GRAYSON ET AL 2003;75: VALVULAR OPERATION AS RISK FACTOR FOR ARF 1835 incidence and risk factors in 2009 consecutive patients. J Thorac Cardiovasc Surg 1993;41: Corwin HL, Sprague SM, DeLaria GA, Norusis MJ. Acute renal failure associated with cardiac operations: a casecontrol study. J Thorac Cardiovasc Surg 1989;98: Mangos GJ, Brown MA, Chan WY, Horton D, Trew P, Whitworth JA. Acute renal failure following cardiac surgery: incidence, outcomes and risk factors. Aust NZ J Med 1995; 25: Frost L, Pedersen RS, Lund O, Hansen OK, Hansen HE. Prognosis and risk factors in acute, dialysis-requiring renal failure after open-heart surgery. Scand J Thorac Cardiovasc Surg 1991;25: Wynne-Jones K, Jackson M, Grotte G, Bridgewater B, on behalf of the north west regional cardiac surgery audit steering group. Limitations of the Parsonnet score for measuring risk stratified mortality in the north west of England. Heart 2000;84: Allison PD. Logistic regression using the SAS system: theory and application. Cary, NC: SAS Institute Inc, Hosmer D, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons Inc, Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143: Abel RM, Buckley MJ, Austen WG, Barnett GO, Beck CH, Fischer JE. Etiology, incidence, and prognosis of renal failure following cardiac operations: results of a prospective analysis of 500 consecutive patients. J Thorac Cardiovasc Surg 1976;71: Suen WS, Mok CK, Chiu SW, et al. Risk factors for development of acute renal failure requiring dialysis in patients undergoing cardiac surgery. Angiology 1998;49: Hilberman M, Myers BD, Carrie BJ, Derby G, Jamison RL, Stinson EB. Acute renal failure following cardiac surgery. J Thorac Cardiovasc Surg 1979;77: Gamoso MG, Phillips-Bute B, Landolfo KP, Newman MF, Stafford-Smith M. Off-pump versus on-pump coronary artery bypass surgery and post-operative renal dysfunction. Anesth Analg 2000;91: Heikkinen L, Harjula A, Merikallio E. Acute renal failure related to open-heart surgery. Ann Chir Gynaecol 1985;74: Llopart T, Lombardi R, Forselledo M, Andrade R. Acute renal failure in open heart surgery. Renal Failure 1997;19: Deal DD, Jones TJ, Hammon JW, Vernon JC, Wall MH, Stump DA. Hypothermic cardiopulmonary bypass in dogs does not protect the kidney from embolization. Anesth Analg 2000;90:S Deal DD, Jones TJ, Vernon JC, Zboyovski JM, Stump DA. Real time OPS imaging of embolic injury of the renal micro-circulation during cardiopulmonary bypass. Anesth Analg 2001;92:S Lemmer JH Jr, Stanford W, Bonney SL, et al. Aprotinin for coronary artery bypass grafting: effect on postoperative renal function. Ann Thorac Surg 1995;59: Lemmer JH Jr, Dilling EW, Morton JR, et al. Aprotinin for primary coronary artery bypass grafting: a multicentre trial of three dose regimens. Ann Thorac Surg 1996;62: The Society of Thoracic Surgeons: Fortieth Annual Meeting Mark your calendars for the Fortieth Annual Meeting of The Society of Thoracic Surgeons, which will be held in San Antonio, Texas, January 26 28, The program will provide in-depth coverage of thoracic surgical topics selected to enhance and broaden the knowledge of practicing thoracic and cardiac surgeons. Traditional abstract presentations as well as topic-specific ancillary sessions and courses will make up the continuing medical education opportunities that will be offered at the Fortieth Annual Meeting. Advance registration forms, hotel reservation forms, and details regarding transportation arrangements, as well as the complete meeting program, will be mailed to Society members. Also, complete meeting information will be available on the Society s Web site located at Nonmembers wishing to receive information on attending the meeting may contact the Society s Secretary, Gordon F. Murray. Abstracts for the meeting must be submitted electronically. The electronic submission form may be accessed at There is no charge for submitting your abstract electronically. The electronic abstract submission deadline is August 18, 2003, at 12 Midnight, CDT. Video submission deadline is August 8, All abstracts must be submitted using the electronic forms located at Any questions may be directed to the STS headquarters. Gordon F. Murray, MD Secretary The Society of Thoracic Surgeons 633 N Saint Clair St, Suite 2320 Chicago, IL Telephone: (312) ; fax: (312) sts@sts.org website: by The Society of Thoracic Surgeons Ann Thorac Surg 2003;75: /03/$30.00 Published by Elsevier Inc