Ann Thorac Surg 2005;80:163-169
© 2005 The Society of Thoracic Surgeons
Original article: Cardiovascular
Effect of the Left Internal Mammary Artery to the Left Anterior Descending Artery on Mortality and Morbidity After Combined Coronary and Valve Operations
Shishir Karthik, FRCSa,
Arun K. Srinivasan, FRCSa,
Antony D. Grayson, BSb,*,
Tim Friede, PhDc,
Brian M. Fabri, FRCS(Ed)a
a Department of Cardiothoracic Surgery, The Cardiothoracic Centre-Liverpool, Liverpool, United Kingdom
b Department of Research and Development, The Cardiothoracic Centre-Liverpool, Liverpool, United Kingdom
c Department of Mathematics and Statistics, Lancaster University, Lancaster, United Kingdom
Accepted for publication January 20, 2005.
* Address reprint requests to Mr Grayson, The Cardiothoracic Centre-Liverpool, Thomas Dr, Liverpool, L14 3PE, United Kingdom (Email: tony.grayson{at}ctc.nhs.uk).
 |
Abstract
|
|---|
BACKGROUND: The effect of using the left internal mammary artery in combined coronary and valve operations have not been fully investigated. We aimed to quantify the impact of the left internal mammary artery to the left anterior descending artery on early and mid-term outcomes in these patients.
METHODS: Data was collected prospectively on 630 consecutive patients who underwent revascularization of the left anterior descending artery with concomitant valve operations between April 1997 and March 2003. Multivariate logistic regression and Cox proportional hazards analyses were used to adjust in-hospital outcomes and Kaplan-Meier survival curves. A propensity score for left internal mammary artery use was constructed to control for selection bias.
RESULTS: The left internal mammary artery was used in 478 (75.9%) patients. Univariate analyses found left internal mammary artery patients had significantly lower in-hospital mortality (6.3% versus 13.2%; p < 0.01) and postoperative renal failure (8.2% versus 13.8%; p = 0.038). After adjusting for treatment selection bias, in-hospital mortality (adjusted odds ratio, 0.77; p = 0.45) and renal failure (adjusted odds ratio, 0.94; p = 0.86) were no longer significantly different. A total of 171 (27.1%) deaths occurred during the follow-up, with a total follow-up of 2,325 patient-years. The crude relative risk for the left internal mammary artery was 0.67 (p = 0.015). After adjusting for the propensity score, the adjusted relative risk was 0.91 (p = 0.62).
CONCLUSIONS: The left internal mammary artery does not adversely affect the short-term and medium-term outcomes in patients undergoing concomitant coronary and valve operations. Survival at 7 years was similar with or without the use of the left internal mammary artery.
|
Introduction
|
|---|
The left internal mammary artery (LIMA) is universally acknowledged as the "gold standard" of surgical revascularization of the left anterior descending coronary artery (LAD) [1]. Since the benefits of LIMA to LAD were demonstrated by Loop and colleagues [2], several other reports have shown the benefits of using the LIMA [3–5]. These have not only confirmed the long-term benefits of using the LIMA, but have revealed a beneficial effect on in-hospital outcomes. Although we have been unable to demonstrate an early beneficial effect of LIMA use in isolated coronary artery bypass grafting (CABG), our previous work has shown a clear medium-term survival advantage in patients receiving LIMA to LAD [6].
In the United Kingdom, patients undergoing combined valve and coronary surgery constitute about 8% of the total workload [7]. During a 10-year period from 1992 to 2001, more than 120,000 combined procedures have been performed in the United States, which represents 6.3% of the adult cardiac surgical work in the same period (STS database executive report 2002: www.ctsnet.org/file/2002ExecutiveReport_Rev.pdf). These patients represent a unique combination of pathologic entities resulting in a wide variety of clinical and hemodynamic lesions, often necessitating complex surgical procedures. These procedures carry a significantly greater risk compared with isolated CABG or isolated valve surgery. Although LIMA usage is routine practice in CABG patients, relatively little is known about the possible impact of LIMA to LAD in patients undergoing combined procedures.
This study evaluates the factors affecting LIMA usage in this subgroup of patients and the impact of LIMA to LAD grafting on in-hospital and medium-term outcomes.
|
Material and Methods
|
|---|
Patient Population and Data
A total of 630 consecutive patients underwent first-time CABG to the LAD with concomitant valve surgery between April 1, 1997, and March 31, 2003, at the Cardiothoracic Centre-Liverpool. Data collection methods and definitions have been previously published [8]. Data were collected prospectively and included preoperative (Table 1), operative (Table 2), and in-hospital outcome (Table 3) variables.
In-hospital mortality 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. Postoperative stroke was defined as a new focal neurologic deficit or comatose state occurring postoperatively that persisted for more than 24 hours after its onset. We excluded confused states, transient cerebral events, and intellectual impairment from our study to avoid any subjective bias. Postoperative myocardial infarction was defined as a new Q wave postoperatively in two or more contiguous leads on an electrocardiogram or a significant rise in postoperative creatine kinase-MB with hemodynamic and echocardiographic signs of myocardial infarction. Postoperative bleeding was defined as bleeding that required surgical reexploration after initial departure from the operating theater. Renal failure was defined as patients with a postoperative creatinine level greater than 200 µmol/L or patients requiring dialysis. Criteria for diagnosing superficial and deep sternal wound infections were in accordance with the guidelines published by the US Centers for Disease Control and Prevention [9].
Patient Follow-Up
Patient records were linked to the National Strategic Tracing Service, which records all-cause mortality in the United Kingdom, to establish current vital status. The cause of death was not available from this system. Patients were matched to the National Strategic Tracing Service on the basis of patient name, National Health Service number, date of birth, sex, and postal code.
Statistical Analysis
Categorical variables are shown as a percentage. Comparisons were made with 
2 tests as appropriate. Standard statistical tests were used to calculate odds ratios and 95% confidence intervals (CI). Parsonnet scores, modified to a regional standard [8], and the logistic version of the European System for Cardiac Operative Risk Evaluation (EuroSCORE) [10] were both calculated to assess differences in patient case mix and are shown as a median with 25th and 75th percentiles. Logistic regression was used to examine the impact of LIMA usage on in-hospital mortality and morbidity, while adjusting for differences in patient characteristics (treatment selection bias) [11]. Deaths occurring as a function of time were described using the product-limit methodology of Kaplan-Meier [12]. To control for treatment selection bias, we used Cox proportional hazards analysis to calculate adjusted relative risk ratios with 95% CI and to perform risk adjustment of the Kaplan-Meier survival curves [13].
Treatment selection bias was controlled for by constructing a propensity score [14]. The propensity score was the probability that a patient would receive LIMA, and was constructed from all the variables listed in Table 1 (C statistic = 0.71). Once the propensity score is constructed for each patient, there are three ways of using the score for comparisons: matching, stratification, and multivariable adjustment. Because of the small sample size available to us for this study, we decided to use multivariable adjustment because matching would have reduced the study size even further and stratification can be difficult to interpret. The propensity score was then included along with the comparison variable (LIMA versus no LIMA) in multivariable analyses of outcome, producing adjusted odds ratios as shown in Table 4. The propensity score adjusted for the treatment selection bias, which was evident in Table 1, between the two groups [14].
Our comparisons were supplemented by repeating all the logistic regression and Cox proportional hazards analyses, but adjusting for the variables listed in Table 1 directly, as apposed to including them into a propensity score. This was because of concerns that some of the confounding variables might have a different influence (weighting) between one outcome and the next, which would be missed by using the propensity score. We also included the year of operation as a confounding variable because of concerns of potential time effects on our results. All the supplementary analyses results were entirely consistent with the original analyses and are not shown.
In all cases a p value less than 0.05 was considered significant. All statistical analysis was performed retrospectively with SAS for Windows Version 8.2 (SAS Institute, Cary, NC).
|
Results
|
|---|
Left Internal Mammary Artery Use
Of the 630 patients who underwent revascularization of the LAD with concomitant valve surgery, 478 (75.9%) received LIMA. The yearly usage of LIMA has increased from 57.5% in the financial year 1997/1998 (April 1, 1997 to March 31, 1998) to 89.5% in the financial year 2002/2003 (p value for trend < 0.001).
Table 1 lists patient characteristics and the rate of LIMA use by various subgroups. Clinical factors associated with low LIMA use included older age, females, poor left ventricular ejection fraction (<30%), emergency surgery, higher New York Heart Association status, and patients with comorbid disease (respiratory disease and renal dysfunction). The median modified Parsonnet score for the LIMA group was 11.2 (25th and 75th percentiles, 7.0 to 15.4) compared with 13.6 (25th and 75th percentiles, 9.8 to 18.9) in the non-LIMA group (p < 0.001). The logistic EuroSCORE was also significantly lower in the LIMA group compared with the non-LIMA group (6.3 [25th and 75th percentiles, 3.7 to 10.9] versus 9.9 [25th and 75th percentiles, 5.9 to 20.1]; p < 0.001).
The type of valve operation performed, the number of grafts used, and duration of cardiopulmonary bypass and aortic cross-clamp for LIMA and non-LIMA patients are shown in Table 2. The type of valve operation was similar between both groups. All aortic valves were replaced. Of the mitral valves operated on, only 60.6% were replaced in the LIMA group compared with 65.8% in the non-LIMA patients (p = 0.56). The rate of LIMA use in single aortic valve (n = 469), single mitral valve (n = 141), and double aortic and mitral valve (n = 15) operations was 75.7%, 76.6%, and 73.3%, respectively.
In-Hospital Outcomes
Crude and adjusted odds ratios for in-hospital outcomes (LIMA versus no LIMA) are shown in Tables 3 and 4, respectively. Hospital mortality was 6.3% in patients receiving the LIMA and 13.2% in the patients not receiving the LIMA (p = 0.006; Table 3). After multivariate adjustment was made with the propensity score, LIMA patients did not have a lower incidence of in-hospital mortality (Table 4).
Table 3 shows that in the univariate analysis, LIMA patients had lower rates of renal failure, reduced need for inotropic support, and a shorter duration of intensive care unit stay. Although not statistically significant, LIMA patients also tended to have a shorter duration of mechanical ventilation (p = 0.077).
After multivariate adjustment with the propensity score, there was no association between LIMA use and postoperative stroke, myocardial infarction, reexploration for bleeding, renal failure, deep and superficial sternal wound infection, need for inotropic and intraaortic balloon pump support, duration of mechanical ventilation, intensive care unit length of stay, and postoperative length of stay (Table 4).
Follow-Up Mortality
One hundred seventy-one deaths (27.1%) occurred during the study period, with a mean follow-up of 3.7 years (standard deviation, 2.1 years). The crude relative risk of long-term mortality for LIMA patients was 0.67 (95% CI, 0.47 to 0.95; p = 0.015; Fig 1). Freedom from death in the LIMA patients at 30 days, and 1, 2, 3, 4, 5, 6, and 7 years was 95.4%, 88.5%, 84.5%, 81.2%, 76.9%, 74.4%, 71.0%, and 63.3%, respectively, compared with 89.5%, 82.2%, 78.9%, 74.1%, 68.5%, 61.8%, 58.2%, and 55.1% for the non-LIMA patients.
View larger version (21K):
[in this window]
[in a new window]
|
Fig 1. Observed survival after combined coronary and valve operations. (LIMA = left internal mammary artery.)
|
|
After multivariate adjustment for the propensity score, the adjusted relative risk of long-term mortality for LIMA patients was 0.91 (95% CI, 0.61 to 1.37; p = 0.62). The adjusted Kaplan-Meier survival curves are shown in Figure 2. The adjusted freedom from death in the LIMA patients at 30 days, and 1, 2, 3, 4, 5, 6, and 7 years was 94.3%, 87.6%, 84.1%, 80.4%, 75.9%, 72.2%, 69.1%, and 63.8%, respectively, compared with 93.8%, 86.5%, 82.7%, 78.8%, 74.0%, 70.0%, 66.7%, and 61.2% for the non-LIMA patients.
View larger version (15K):
[in this window]
[in a new window]
|
Fig 2. Adjusted survival after combined coronary and valve operations, adjusted for the propensity score. The propensity score included the following factors: age, sex, body mass index, New York Heart Association class, respiratory disease, diabetes, cerebrovascular disease, renal dysfunction, ejection fraction, extent of coronary disease, aortic valve gradient, systolic pulmonary artery pressure, and emergency status. (LIMA = left internal mammary artery.)
|
|
Further analysis of patients with combined aortic valve and CABG surgery found that the crude relative risk of long-term mortality for LIMA patients was 0.80 (95% CI, 0.55 to 0.1.17; p = 0.25). Adjustment with the propensity score gave an adjusted relative risk of long-term mortality for LIMA patients of 1.05 (95% CI, 0.69 to 1.60; p = 0.82). Analysis of patients with combined mitral valve and CABG surgery revealed that the crude relative risk of long-term mortality for LIMA patients was 0.44 (95% CI, 0.23 to 0.84; p = 0.013). After adjusting for the propensity score, the adjusted relative risk of long-term mortality for LIMA patients was no longer significantly different at 0.56 (95% CI, 0.26 to 1.21; p = 0.14).
|
Comment
|
|---|
The advantages of LIMA to LAD and its beneficial effects on long-term survival are established. Despite this, the usage of LIMA to LAD varies among institutions and surgeons. Cleveland and associates [15] have mentioned an 81% LIMA usage rate across all coronary revascularizations in the United States. In a recent study by Mack and coworkers [16], about 15% of the patients undergoing CABG in 69 hospitals across 17 states in the United States between 1999 and 2002 did not receive any arterial grafts. Our experience in the northwest of England reveals a widespread variation in practice with LIMA to LAD usage varying among individual surgeons from as low as 61% to as high as 97% (see www.nwheartaudit.nhs.uk). At our center there has been a rise in LIMA usage in isolated CABG, and during the last 3 years we have achieved a 95% LIMA to LAD rate in patients with significant LAD disease.
There is much less information about LIMA usage in combined procedures. The numbers have varied between 0% and 36% in different series [17–19]. More recently, Gall and colleagues [20] reported an LIMA to LAD grafting rate of nearly 60% (135 of 227 patients) in patients undergoing combined aortic valve and CABG. In our study, 75.7% of patients undergoing combined aortic valve and CABG received LIMA. We have shown a significant rise in LIMA usage during this 6-year period from 57.5% to 89.5%, with an overall LIMA usage of 75.9% in all combined CABG and valve operations. Although there has been an increased usage of LIMA in combined procedures, it is still much lower than for isolated CABG.
Not surprisingly, the patients with a lower rate of LIMA usage in the combined valve and coronary operations, ie, elderly patients (age > 70 years), women, patients with preexisting respiratory and renal dysfunction, patients with poor left ventricular ejection fraction (<30%), worse New York Heart Association status, and emergency surgery, are similar to patient groups well recognized as having a lower rate of LIMA usage in patients undergoing isolated CABG [5, 6]. Concurrent valve surgery tends to reduce the levels of LIMA usage even further in these patients.
There are other potential reasons for why LIMA was not used within our study. These could include technical issues such as injury to the LIMA while harvesting or poor flow. Our previous work looking into the reasons for not using the LIMA in isolated CABG found that 2% of cases did not have the LIMA used because of poor flow or damage to the LIMA [6]. Unfortunately the recording of reasons for not using the LIMA in the patients’ medical records for combined cases was poorly filled in and deemed unusable.
The significant short-term and long-term benefits, both in terms of overall survival and patency rates of LIMA to LAD grafting, is well documented for CABG in several studies [2–5, 21], but has not really been well established in combined procedures. In this present study, LIMA usage seems to be associated with a lower in-hospital mortality and renal failure, a reduced need for inotropic agents, and shorter intensive care unit stays as well as a better survival for the medium term. However, we believe that this is a reflection of the differences in patient characteristics between the two groups, as all these results lose their significance on multivariate adjustment.
Although we have shown that LIMA usage is not associated with any increase in adverse short-term outcomes, we have been unable to demonstrate any benefit of using LIMA in this select group of patients for the medium term. It is most likely that the valvular component of these combined procedures has a very significant impact on survival, which may overwhelm any benefits that LIMA usage is likely to confer. In coronary surgery, long-term studies clearly demonstrate that there is a significant difference in the patency rates, event-free survival, and overall survival between patients receiving LIMA and those receiving venous grafts. After an initial attrition in vein graft patency during the first 12 months, LIMA and venous graft patency closely mirror each other for about 5 years followed by a further accelerated rate in vein graft occlusion as one reaches 10- and 15-year follow-up. It is now fairly well established that the patency rates of LIMA are about 90% for 10-year follow-up; hence, the differences between the two groups are likely to widen. Longer follow-up may therefore show that LIMA usage has a bigger impact on survival in patients undergoing combined procedures.
Another reason for why we are unable to demonstrate a potential benefit from using the LIMA in combined cases may be related to the fact that these patients have less severe coronary disease than patients receiving isolated CABG, and LIMA use theoretically may not have an impact on patients with less than critical coronary disease in the LAD. Only about half of the study group had triple-vessel disease (stenosis greater than 70% in all three vessels). A subgroup analysis of these patients, therefore guaranteeing that the LAD had critical stenosis, is inconclusive, however, because of the small sample size available.
Gall and associates [20] were also able to demonstrate the safety of LIMA usage in patients undergoing combined procedures, but went one step further and suggested a survival benefit when using the LIMA. Their study included a total of 227 patients undergoing CABG with aortic valve replacement or repair, 92 of whom did not receive the LIMA. In our study, we showed no survival benefit in the 469 patients undergoing combined aortic valve and CABG, or even any other combined CABG and valve operation. The follow-up period in this present study was on average 3.7 years, which is similar to that of Gall and colleagues [20] with an average of 3.4 years. A possible reason why they showed a survival benefit, and our study did not, is that although they did use Cox proportional hazards analysis to identify LIMA as a beneficial factor for long-term survival, certain patient characteristics were not adjusted for. The non-LIMA patients were a much higher-risk group. Factors such as age and obesity, although not associated with long-term mortality in their analyses, were significantly different between LIMA and non-LIMA patients and therefore could contribute to systematic bias [22]. Adjusting for a propensity score or balancing score, as we have done, not only adjusts for significant factors but augments them with other variables, even if not significant. The aim is to balance the patient characteristics in both study groups by incorporating everything that may relate to potential systematic bias [14].
There are some limitations, which may affect the conclusions drawn from this report. These include the retrospective, observational design of the study. An important limitation is the fact that we have only been able to adjust our results for known confounders, but there may be hidden confounders still affecting our inferences. We have used propensity score adjustment to account for the known selection bias. A further limitation is that we have only assessed all-cause mortality rates, which means that cardiac-related deaths are potentially overpredicted. However, we do not expect this to significantly affect our conclusions, as the chances of a noncardiac-related death are equal in both groups. This study also does not assess other medium-term to long-term outcomes for these patients, eg, graft patency, reintervention rates, and quality of life, data that were not available to us. A final limitation is the potential for type 2 errors caused by the small study population and in some cases small incidences of in-hospital outcomes. This is of particular importance when assessing the impact of LIMA on deep sternal wound infections (only 15 recorded), with an adjusted odds ratio of 2.63 (95% CI, 0.52 to 13.19).
In conclusion, we have shown that LIMA usage in patients undergoing combined procedures is not associated with any increased risk of adverse events and therefore should be continued to be used in these patients. Further follow-up in this cohort of patients during the next 5 years will be of great interest, as any relative survival benefit of LIMA use may become apparent.
|
Notice From the American Board of Thoracic Surgery
|
|---|
The 2005 Part I (written) examination will be held on Monday, December 5, 2005. It is planned that the examination will be given at multiple sites throughout the United States using an electronic format. The closing date for registration is August 1, 2005. Those wishing to be considered for examination must apply online at www.abts.org.
To be admissible to the Part II (oral) examination, a candidate must have successfully completed the Part I (written) examination.
A candidate applying for admission to the certifying examination must fulfill all the requirements of the Board in force at the time the application is received.
Please address all communications to the American Board of Thoracic Surgery, 6333 N St. Clair St, Suite 2320, Chicago, IL 60611; telephone: (312) 202-5900; fax: (312) 202-5960; e-mail: mailto:info{at}abts.org.
|
Acknowledgments
|
|---|
We would like to acknowledge the cooperation given to us by all the consultant cardiac surgeons at the Cardiothoracic Centre-Liverpool: John A. C. Chalmers, Walid C. Dihmis, Brian M. Fabri, Elaine M. Griffiths, Neeraj K. Mediratta, Richard D. Page, D. Mark Pullan, Abbas Rashid, and W. Ian Weir. We would also like to thank Janet Deane, who maintains the quality and ensures completeness of data collected in our Cardiac Surgery Registry.
|
References
|
|---|
- Cohn LH. Use of internal mammary artery graft and in-hospital mortality and other adverse outcomes associated with coronary artery bypass surgery Circulation 2001;103:483-484.[Free Full Text]
- Loop FD, Lytle BW, Cosgrove DM, et al. Influence of internal mammary artery graft on 10-year survival and other cardiac events N Engl J Med 1986;314:1-6.[Abstract]
- Grover FL, Johnson RR, Marshall G, Hammermeister KE. Impact of mammary grafts on coronary bypass operative mortality and morbidity Ann Thorac Surg 1994;57:559-569.[Abstract]
- Cameron AA, Green GE, Brogno DA, et al. Internal thoracic artery grafts20-year clinical follow-up. J Am Coll Cardiol 1995;25:188-192.[Abstract]
- Leavitt BJ, O’Connor GT, Olmstead EM, et al. Use of internal mammary artery graft and in-hospital mortality and other adverse outcomes associated with coronary artery bypass surgery Circulation 2001;103:507-512.[Abstract/Free Full Text]
- Karthik S, Srinivasan AK, Grayson AD, Jackson M, Mediratta NK. Left internal mammary artery to the left anterior descending arteryeffect on morbidity and mortality and reasons for non-usage. Ann Thorac Surg 2004;78:142-148.[Abstract/Free Full Text]
- The Society of Cardiothoracic Surgeons of Great Britain and Ireland National adult cardiac surgical database report 2000–2001. Oxfordshire: Dendrite Clinical Systems; 2002October.
- Wynne-Jones K, Jackson M, Grotte G, Bridgewater B. On behalf of the north west regional cardiac surgery audit steering groupLimitations of the Parsonnet score for measuring risk stratified mortality in the north west of England. Heart 2000;84:71-78.[Abstract/Free Full Text]
- Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. The Hospital Infection Control Practices Advisory CommitteeGuideline for prevention of surgical site infection, 1999. Infect Control Hosp Epidemiol 1999;20:247-278.
- Michel P, Roques F, Nashef SAM, EuroSCORE Project Group Logistic or additive EuroSCORE for high-risk patients? Eur J Cardiothorac Surg 2003;23:684-687.[Abstract/Free Full Text]
- Hosmer D, Lemeshow S. Applied logistic regressionNew York, NY: John Wiley & Sons; 1989.
- Kaplan EL, Meier P. Nonparametric estimation from incomplete observations J Am Stat Assoc 1958;53:547-581.
- Cox DR. Regression models and life-tables J R Stat Soc 1972;34:187-220.
- Blackstone E. Comparing apples and oranges J Thorac Cardiovasc Surg 2002;123:8-15.[Free Full Text]
- Cleveland Jr JC, Shroyer ALW, Chen AY, Peterson E, Grover FL. Off-pump coronary artery bypass grafting decreases risk-adjusted mortality and morbidity Ann Thorac Surg 2001;72:1282-1289.[Abstract/Free Full Text]
- Mack MJ, Brown PP, Kugelmass AD, et al. Current status and outcomes of coronary revascularization 1999 to 2002148,396 surgical and percutaneous procedures. Ann Thorac Surg 2004;77:761-768.[Abstract/Free Full Text]
- Akins CW, Buckley MJ, Daggett WM, Hilgenberg AD, Austen GW. Myocardial revascularisation with combined aortic and mitral valve replacements J Thorac Cardiovasc Surg 1985;90:272-277.[Abstract]
- Flameng W, Szecsi J, Sergeant P, Daenen W, Herijgers P, Scheys I. Combined valve and coronary artery bypass surgeryearly and late results. Eur J Cardiothorac Surg 1994;8:410-419.[Abstract]
- Bauer EP, Schonburg M, Schwarz T, Piepho A, Klovekorn WP. Valve surgery combined with coronary artery operationis the use of internal mammary artery a predictor for early complications. Eur J Cardiothorac Surg 1996;10:248-252.[Abstract]
- Gall Jr S, Lowe JE, Wolfe WG, Oldham Jr HN, Trigt III PV, Glower DD. Efficacy of the internal mammary artery in combined aortic valve replacement-coronary artery bypass grafting Ann Thorac Surg 2000;69:524-530.[Abstract/Free Full Text]
- Tatoulis J, Buxton BF, Fuller JA. Patencies of 2,127 arterial to coronary conduits over 15 years Ann Thorac Surg 2004;77:93-101.[Abstract/Free Full Text]
- Rosenbaum PR. Optimal matching for observational studies J Am Stat Assoc 1989;84:1024-1032.
Top
Abstract
Introduction
