ATS
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to Personal Folders
Right arrow Download to citation manager
Right arrow Author home page(s):
Mitchell J. Magee
Todd M. Dewey
James R. Edgerton
William H. Ryan
Syma Prince
Michael J. Mack
Right arrow Permission Requests
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Magee, M. J.
Right arrow Articles by Mack, M. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Magee, M. J.
Right arrow Articles by Mack, M. J.
Related Collections
Right arrow Electrophysiology - arrhythmias

Ann Thorac Surg 2007;83:1707-1712
© 2007 The Society of Thoracic Surgeons

Original Articles: Cardiovascular

Atrial Fibrillation After Coronary Artery Bypass Grafting Surgery: Development of a Predictive Risk Algorithm

Mitchell J. Magee, MDa,b, Morley A. Herbert, PhDa,*, Todd M. Dewey, MDa,b, James R. Edgerton, MDa,b, William H. Ryan, MDb,c, Syma Prince, RN, BSNb, Michael J. Mack, MDa,b

a Medical City Dallas Hospital, Dallas, Texas
b Cardiopulmonary Research Science and Technology Institute, Dallas, Texas
c Presbyterian Hospital of Dallas, Dallas, Texas

Accepted for publication December 18, 2006.

* Address correspondence to Dr Herbert, Medical City Dallas Hospital, 7777 Forest Lane, Suite C-740, Dallas, TX 75230 (Email: herbert{at}healthcare.com).

Presented at the Fifty-second Annual Meeting of the Southern Thoracic Surgical Association, Orlando FL, Nov 10–12, 2005.


   Abstract
Top
 Abstract
Introduction
 Patients and Methods
 Results
 Comment
 Discussion
 References
 
Background: Atrial fibrillation is a costly complication occurring in 15% to 40% of patients after coronary artery bypass grafting (CABG). Aggressive prophylactic treatment should be directed toward and limited to selected high-risk patients. Utilizing perioperative risk factors, we sought to develop an algorithm to predict the relative risk of developing postoperative atrial fibrillation in patients undergoing CABG.

Methods: Data were extracted from our Society of Thoracic Surgeons Database on 19,620 patients undergoing CABG between January 1995 and July 2006. We used perioperative risk factors to develop a logistic regression equation predictive for the development of postoperative atrial fibrillation. A total of 19,083 patients had complete data and were used to construct the final model. The model was used to compare the predicted probability of atrial fibrillation with the known outcome in the patients divided into deciles by probability. Bootstrap procedures were used to determine the confidence limits of the ß coefficients.

Results: A regression model was developed with 14 significant indicators. Those showing the greatest predictive influence included the patient age, the need for prolonged ventilation (24 hours or more), the use of cardiopulmonary bypass, and preoperative arrhythmias. The model showed acceptable concordance between observed and predicted (72.3%), a receiver operating characteristic curve area of 0.72, and Hosmer-Lemeshow probability of 0.19. When applied to the patient population, the calculated risk in those who did not develop AF was 0.179 ± 0.116 and for those with AF, 0.284 ± 0.153 (p < 0.001).

Conclusions: A validated predictive risk algorithm for developing postoperative atrial fibrillation can reliably stratify patients undergoing CABG into high-risk and low-risk groups. This may be used preoperatively to appropriately target high-risk patients for aggressive prophylactic treatment.


   Introduction
Top
 Abstract
 Introduction
Patients and Methods
 Results
 Comment
 Discussion
 References
 
More than 600,000 cardiac surgical procedures are performed annually in the United States, of which more than 400,000 are coronary artery bypass grafting (CABG) operations [1]. Atrial fibrillation is one of the most common complications occurring after cardiac surgery, with as many as 10% to 40% of all patients undergoing CABG experiencing new onset postoperative atrial fibrillation [2–9].

Owing to advances in surgery, anesthesia, and postoperative care—and despite a trend for patients undergoing these procedures to be of higher risk—operative mortality and morbidity remain low and, in fact, have declined in recent years [10, 11]. Despite this general decline in complications, the incidence of postoperative atrial fibrillation has not decreased and has actually appeared to be increasing, most likely attributable to the increasing proportions of CABG procedures performed in elderly patients [6, 8].

Although commonly regarded as a benign, self-limiting complication not associated with increased mortality, atrial fibrillation may result in significant morbidity, including hypertension, palpitations, pain, fatigue, dyspnea, or generalized anxiety. Postoperative atrial fibrillation is also associated with congestive heart failure, renal insufficiency, prolonged ventilation, readmission to the intensive care unit, and a threefold to fourfold increased risk of early postoperative stroke [6, 8]. Studies have shown that postoperative atrial fibrillation resulted in longer intensive care unit (ICU) stays, an increased risk of readmission to the ICU, prolonged hospital stay, and increased hospital costs [3, 8, 12]. Atrial fibrillation is also the leading cause of hospital readmission after hospital discharge following cardiac surgery [13].

Because of the enormous clinical and economic impact of this complication, numerous attempts have been made to identify risk factors for postoperative atrial fibrillation in an effort to not only provide insight into pathophysiology but also allow better assessment of prophylactic treatment strategies by directing treatment at those patients most likely to benefit. Identifying a population of patients who are at increased risk for developing postoperative atrial fibrillation would allow for a more targeted preventative or therapeutic intervention strategy that would reduce the potential for antiarrhythmic-related toxicity and decrease the overall cost of prophylactic treatment [14, 15]. Towards this goal, we sought to develop a risk algorithm by using readily available preoperative, intraoperative, and early postoperative indicators for predicting postoperative atrial fibrillation in patients undergoing CABG.


   Patients and Methods
Top
 Abstract
 Introduction
 Patients and Methods
Results
 Comment
 Discussion
 References
 
Patients
The Cardiopulmonary Research Science and Technology Institute (CRSTI) prospectively collects data on all patients undergoing cardiac operations for a group practice of 22 surgeons practicing in 18 hospitals. Data collection by various clinical professionals is initiated at the time of hospital admission, maintained during hospitalization, and completed at the first postoperative visit. Patient data are collected and maintained according to The Society of Thoracic Surgeons (STS) National Cardiac Surgery Database guidelines and definitions (available at: http://www.ctsnet.org/file/rptDataSpecifications252_1_ForVendorsPGS.pdf.). The data are stored in a customized STS computerized database, and a subset of the data is audited twice yearly for every surgeon’s practice. The current year’s data set is exported to the Society of Thoracic Surgeons (STS) National Adult Cardiac Database semi-annually [16].

A retrospective review of data collected by CRSTI on all patients undergoing cardiac operations between January 1, 1995, and July 31, 2006, identified 19,620 patients that underwent isolated CABG surgery; of these, 19,083 had complete data that contributed to the model building. The North Texas Institutional Review Board at Medical City Dallas Hospital approved the study. Patients with postoperative atrial fibrillation were determined from the selected database field (Comps-Other-A-Fib) defined as "patient had a new onset of atrial fibrillation/flutter (AF) requiring treatment. Does not include recurrence of AF which had been present preoperatively."

Patients with a preoperative history of atrial dysrhythmias, defined as atrial fibrillation/flutter requiring treatment within 2 weeks before surgery, were excluded to avoid confusing the incidence of postoperative atrial fibrillation with recurrence of their preexisting condition. Patients with a postoperative length of stay of less than 2 days were also excluded owing to an inability to accurately determine the incidence of postoperative atrial fibrillation in this group. Changes in the STS database definitions during the period studied required that some fields that are now ordinal be redefined as a binary variable so that the data could be appropriately matched.

Statistical Analysis
Patients were grouped by the presence or absence of postoperative atrial fibrillation, and all preoperative as well as selected intraoperative and postoperative variables were compared between groups. Categorical variables were tested using either the {chi}2 or the Fisher exact test, where appropriate, and continuous variables were tested with the two-tailed Student t test using the statistical software SAS 9.1.3 (SAS Institute, Cary, NC). Values of p ≤ 0.05 were considered significant. Those variables demonstrating statistical significance between the groups by univariate analysis were used as a starting point in building the predictive model (Table 1), recognizing that some of these variables would ultimately be not significant in the final model when controlling for other variables.


View this table:
[in this window]
[in a new window]

  		Table 1 Univariate Analysis of Patients With and Without Postoperative Atrial Fibrillation
 
Selected variables thought to be clinically significant in the development of postoperative atrial fibrillation were also added in as independent variables in a logistic model with atrial fibrillation as the dependent variable. The model building was repeated, eliminating variables that were statistically insignificant and with the smallest coefficients (and odds ratios), while testing different combinations for their effect and ability to improve the area under the curve or the predictive ability. The final model incorporated data from the 19,083 patients who had values for all the fields used in the model.

The calculated probabilities were then divided into deciles between 0 and 1, and the number of patients predicted to have postoperative atrial fibrillation within each decile was compared with the observed incidence. A bootstrap procedure was run 1000 times, and 95% confidence limits for the coefficients calculated. Finally, probabilities were calculated for all records in the data set, and the predicted was compared with the observed.


   Results
Top
 Abstract
 Introduction
 Patients and Methods
 Results
Comment
 Discussion
 References
 
New onset postoperative atrial fibrillation developed in 4215 (21.5%) of the 19,620 total isolated CABG patients. Using univariate analysis, we compared the incidence of different variables in patients with and without atrial fibrillation (Table 1). The incidence of diabetes was not different between patients with and without postoperative atrial fibrillation. Postoperative atrial fibrillation developed in patients who were older, had a lower ejection fraction, had a larger number of distal anastomoses, and who were ventilated postoperatively nearly three times longer. They also had a higher STS Predicted Risk of Mortality score, indicating that they were sicker patients.

The final predictive model had 14 variables that provided statistically significant coefficients and odds ratios. Table 2 lists the variables and their ß coefficients, and the odds ratios are given in Table 3. The results of the bootstrap analysis are shown as 95% confidence intervals for the coefficients. For the final model, the receiver operating characteristic curve (ROC), or c statistic, was 0.72, and the Hosmer-Lemeshow test had a {chi}2 value of 11.2 (p = 0.19), showing a good fit for the model.


View this table:
[in this window]
[in a new window]

  		Table 2 ß-Coefficients and 95% Confidence Intervals for Variables in Atrial Fibrillation Risk Algorithm a
 

View this table:
[in this window]
[in a new window]

  		Table 3 Odds Ratios and 95% Confidence Intervals for the ß Coefficients
 
When the model variables were applied to the data set and the groups were compared, the predicted risk (mean ± standard deviation) for postoperative new atrial fibrillation developing in patients without postoperative atrial fibrillation was 0.179 ± 0.116, and in the group of patients with postoperative atrial fibrillation, it was 0.284 ± 0.153 (p < 0.001). For all patients, the median value was 0.169 (range, 0.001 to 0.965).

Use of the algorithm to decide on initiating treatment before atrial fibrillation starts requires that the patients be divided into high-risk and low-risk groups according to a cutoff value of the calculated probability. High-risk patients could begin prophylactic treatment immediately after surgery. Table 4 presents the effect of varying this breakpoint. Raising the value makes the treatment of the high-risk group more selective, but at the price of placing more atrial fibrillation cases in the low-risk group. Lowering the breakpoint to the median increases the number of patients treated in the high-risk group, but the number of expected cases of atrial fibrillation in the low-risk group drops to 10%.


View this table:
[in this window]
[in a new window]

  		Table 4 Atrial Fibrillation Rates in Low-Risk and High-Risk Groups With Various Breakpoints
 
Figure 1 demonstrates the correlation between the predicted and observed rates expressed in percentages. The predicted rates have been grouped into deciles with the data plotted at the center of the interval. The line indicates a 1:1 relationship. The points accurately track the line, although at higher-risk probabilities it shows increased deviation from the line, partly owing to smaller numbers of patients in those groups.


Figure 1
View larger version (11K):
[in this window]
[in a new window]

  		Fig 1. Correlation between predicted and observed incidence of postoperative atrial fibrillation (AF). The predicted rates have been grouped into deciles with the data plotted at the center of the interval. The line indicates a 1:1 relationship.

 

   Comment
Top
 Abstract
 Introduction
 Patients and Methods
 Results
 Comment
Discussion
 References
 
The development of atrial fibrillation after cardiac surgery is an extremely important and costly complication. With occurrence rates of 10% to 40%, it affects a large patient population and leads to large incremental costs of treatment. In our study, the mean length of stay after bypass surgery increases from 6.1 ± 4.8 days (median, 5 days) for patients without atrial fibrillation to 9.7 ± 8.2 days (median, 7 days) for those developing atrial fibrillation. The ability to determine perioperatively the probability of postoperative atrial fibrillation developing in each patient, and initiate effective prophylactic treatment in patients calculated to be at increased risk, could yield major benefits in terms of morbidity and cost.

It is our intent to clinically validate the algorithm by selecting a threshold probability, above which all patients will be treated prophylactically to prevent atrial fibrillation. Patients admitted for surgery will have their probability calculated, with prolonged ventilation scored as a "No." If the calculated probability is above the threshold, prophylactic treatment is initiated at the end of the operation; otherwise, the patients will not be treated unless the development of prolonged ventilation drives the probability above the threshold value, at which time prophylactic treatment would be initiated.

A major problem in developing the model was that the STS definition of preoperative arrhythmia, which includes atrial fibrillation, was based on short-term observation. In the data definitions used before 2004 (section F, pages 36–7, version 2.41; http://www.ctsnet.org/file/241DataSpecs.pdf), the field only asked about the 2-week period before the patient’s cardiac surgery:

Arrhythmia

Is there a preoperative arrhythmia present within two weeks of the procedure, by clinical documentation of any one of the following:

Atrial fibrillation/flutter requiring Rx; etc, with the child field containing the type of arrhythmia.

Arrhythmia Type

Which arrhythmia is present within two weeks of the procedure; choose one: Atrial fibrillation/flutter requiring Rx. etc.

Since 2004, version 2.52 of the definition (Section F, page 30–1, http://www.ctsnet.org/file/rptDataSpecifications252_1_ForVendorsPGS.pdf) has been broader, asking about the reported history but still limiting the arrhythmia type data to a 2-week period. Thus, patients with paroxysmal atrial fibrillation could have a "yes" to arrhythmia but "none" to the type if they did not have an occurrence within the 2-week window.

Arrhythmia

Indicate whether there is a history of preoperative arrhythmia (sustained ventricular tachycardia, ventricular fibrillation, atrial fibrillation, atrial flutter, third degree heart block) that has been clinically documented or treated with any of the following treatment modalities:

1 ablation therapy
2 AICD
3 pacemaker
4 pharmacological treatment
5 electrocardioversion

Arrhythmia Type

Indicate which arrhythmia is present within two weeks of the procedure; choose one:

• Sustained ventricular tachycardia or ventricular fibrillation requiring cardioversion and/or IV amiodarone
• Third degree heart block
• Atrial fibrillation/flutter requiring Rx
• None

We attempted to eliminate patients who had any record of atrial fibrillation preoperatively, but the collected data might not have indicated all the patients. Because the prevalence in the population is estimated at 0.95% [17], the number of patients with preoperative atrial fibrillation who were inadvertently included might not have been large. We were unable to estimate the number accurately, however.

Prospectively identifying patients at increased risk for developing postoperative atrial fibrillation should allow a more accurate assessment of preventive interventions, focus treatment on those most likely to benefit, minimize unnecessary treatment, and maximize treatment value. To that end, a weighted variable algorithm consisting of 14 readily obtainable clinical indicators was created to determine the predicted risk of postoperative atrial fibrillation developing in patients undergoing CABG. The model was derived from a large population and demonstrates acceptable accuracy and concordance, good selectivity (ROC = 0.722), and calibration (Hosmer-Lemeshow {chi}2 = 9.8, p = 0.28). Prospective validation in other populations is warranted and planned. [14, 15]


   Discussion
Top
 Abstract
 Introduction
 Patients and Methods
 Results
 Comment
 Discussion
References
 
DR ROBERT A. GUYTON (Atlanta, GA): I was struck by the presence of digoxin use ahead of beta-blocker use and ahead of ACE inhibitor use in your algorithm. Was preoperative digoxin use associated with the development of A-fib post-op or with the nondevelopment of A-fib post-op? I couldn’t tell that from your slide.

DR MAGEE: I have to go back and look at that myself, because that was not a very strong factor. The use of digoxin did correlate with the development of AF. It was the use of digoxin that was associated with the development of AF.

DR GUYTON: And it was twice as strong as preoperative beta-blocker use, and in the case of beta-blocker use, preoperative beta-blocker use was associated with non-development or with development of A-fib?

DR MAGEE: It was associated with the development of AF.

DR TARA KARAMLOU (Toronto, Ontario, Canada): I also thought your statistical analysis was excellent. I just wonder with the trend toward developing minimum data sets with accurate discrimination whether you took into account correlation or redundancy? Many of your variables are probably not in fact independent, and I wonder if a more parsimonious model could be created without loss of predictive ability by removing some of these terms that are highly correlated?

DR MAGEE: I am not sure I can answer that. We did test the model in multiple ways. We used stepwise regression and we did remodel the model multiple times to try to find the best fit, and it was a challenge, because there is a lot of interaction, and you can see that the univariate analysis factors did not necessarily correlate in the end with the multivariate analysis. We did try to include those factors that we thought clinically might have significance even though they didn’t necessarily fall out statistically in the univariate analysis.

DR WILLIAM A. BAUMGARTNER (Baltimore, MD): Great presentation. What is your prophylactic regimen, because you said using it you would expect a 60% reduction? Would you mind sharing that with us?

DR MAGEE: That was an assumption basically and we have not validated this prospectively with a treatment regimen. That was based on the literature. We used IV amiodarone or PO amiodarone, and there are a variety of studies that have been done in the literature, and really that is the only one that has been prospectively validated to be beneficial in decreasing length of stay and decreasing the postoperative incidence of atrial fibrillation. The ranges in the literature are anywhere from 7% up to 74% effectiveness in using that regimen, and it is certainly higher if you use postoperative beta-blockers. What we propose and what we intend to do prospectively is to treat all patients with amiodarone. And apparently, again from my reading of the literature, it doesn’t matter whether you treat them right after surgery or within the first 24 hours after surgery, because one of our factors is prolonged ventilation, and so that may pull some people in at that 20% level or above by staying on the vent, and we would plan to treat those patients at that point.


   References
Top
 Abstract
 Introduction
 Patients and Methods
 Results
 Comment
 Discussion
 References
 

  1. American Heart Association. Heart disease and stroke statistics–2005 update. Available at: www.americanheart.org. Accessed Nov 11, 2005.
  2. Ommen SR, Odell JA, Stanton MS. Atrial arrhythmias after cardiothoracic surgery N Engl J Med 1997;336:1429-1434.[Free Full Text]
  3. Aranki SF, Shaw DP, Adams DH, et al. Predictors of atrial fibrillation after coronary artery surgery: current trends and impact on hospital resources Circulation 1996;94:390-397.[Abstract/Free Full Text]
  4. Caretta Q, Mercanti CA, DeNardo D, et al. Ventricular conduction defects and atrial fibrillation after coronary artery bypass grafting: Multivariate analysis of preoperative, intra-operative and postoperative variables Eur Heart J 1991;12:1107-1111.[Abstract/Free Full Text]
  5. Crosby LH, Pifalo WB, Woll KR, et al. Risk factors for atrial fibrillation after coronary artery bypass grafting Am J Cardiol 1990;66:1520-1522.[Medline]
  6. Creswell LL, Schuessler RB, Rosenbloom M, Cox JL. Hazards of postoperative atrial arrhythmias Ann Thorac Surg 1993;56:539-549.[Abstract]
  7. Leitch JW, Thomson D, Baird DK, et al. The importance of age as a predictor of atrial fibrillation and flutter after coronary artery bypass grafting J Thorac Cardiovasc Surg 1990;100:338-342.[Abstract]
  8. Mathew JP, Parks R, Savino JS, et al. Atrial fibrillation following coronary artery bypass surgery: Predictors, outcomes, and resource utilizationMulticenter Study of Perioperative Ischemia Research Group. JAMA 1996;276:300-306.[Abstract]
  9. Hogue Jr CW, Hyder ML. Atrial fibrillation after cardiac operation: risks mechanisms, and treatment Ann Thorac Surg 2000;69:300-306.[Abstract/Free Full Text]
  10. Ivanov J, Weisel RD, David TE, Naylor CD. Fifteen-year trends in risk severity and operative mortality in elderly patients undergoing coronary artery bypass graft surgery Circulation 1998;97:673-680.[Abstract/Free Full Text]
  11. Ferguson Jr TB, Hammill BG, Peterson ED, et al. A decade of change–risk profiles and outcomes for isolated coronary artery bypass grafting procedures, 1990–1999: a report from the STS National Database committee and the Duke Clinical Research Institute Ann Thorac Surg 2002;73:480-490.[Abstract/Free Full Text]
  12. Taylor GJ, Mikell FL, Moses W, et al. Determinants of hospital charges for coronary artery bypass surgery: the economic consequences of postoperative complications Am J Cardiol 1990;65:309-313.[Medline]
  13. Lahey SJ, Campos CT, Jennings B, et al. Hospital readmission after cardiac surgery: does "fast track" cardiac surgery result in cost saving or cost shifting? Circulation 1998;98:II-35-II-40.[Medline]
  14. Hogue Jr CW, Creswell LL, Gutterman DD, et al. Epidemiology, mechanisms and risksAmerican College of Chest Physicians Guidelines for the Prevention and Management of Postoperative Atrial Fibrillation After Cardiac Surgery. Chest 2005;128:9S-16S.[Medline]
  15. Amar D, Weiji S, Hogue Jr CW, et al. Clinical prediction rule for atrial fibrillation after coronary artery bypass grafting J Am Coll Cardiol 2004;44:1248-1253.[Abstract/Free Full Text]
  16. Herbert MA, Prince SL, Williams JL, Magee MJ, Mack MJ. "Are unaudited records from an outcomes database accurate?" Ann Thoracic Surgery 2004;77:1960-1965.[Abstract/Free Full Text]
  17. Go AS, Hylek EM, Phillips KA, et al. Prevalence of diagnosed atrial fibrillation in adults JAMA 2001;285:2370-2375.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 EuropaceHome page
M. Banach, G. Mariscalco, M. Ugurlucan, D. P. Mikhailidis, M. Barylski, and J. Rysz
The significance of preoperative atrial fibrillation in patients undergoing cardiac surgery: preoperative atrial fibrillation--still underestimated opponent
Europace, November 1, 2008; 10(11): 1266 - 1270.
[Abstract] [Full Text] [PDF]

Home page
 J. Thorac. Cardiovasc. Surg.Home page
M. Banach, A. Goch, M. Misztal, J. Rysz, R. Jaszewski, and J. H. Goch
Predictors of paroxysmal atrial fibrillation in patients undergoing aortic valve replacement.
J. Thorac. Cardiovasc. Surg., December 1, 2007; 134(6): 1569 - 1576.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to Personal Folders
Right arrow Download to citation manager
Right arrow Author home page(s):
Mitchell J. Magee
Todd M. Dewey
James R. Edgerton
William H. Ryan
Syma Prince
Michael J. Mack
Right arrow Permission Requests
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Magee, M. J.
Right arrow Articles by Mack, M. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Magee, M. J.
Right arrow Articles by Mack, M. J.
Related Collections
Right arrow Electrophysiology - arrhythmias

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
ANN THORAC SURG ASIAN CARDIOVASC THORAC ANN EUR J CARDIOTHORAC SURG
J THORAC CARDIOVASC SURG ICVTS ALL CTSNet JOURNALS