New-Onset Postoperative Atrial Fibrillation and Long-Term Survival After Aortic Valve Replacement Surgery

New-Onset Postoperative Atrial Fibrillation and Long-Term Survival After Aortic Valve Replacement Surgery Giovanni Filardo, PhD, MPH, Cody Hamilton, PhD, Baron Hamman, MD, Robert F. Hebeler, Jr, MD, John Adams, MPH, and Paul Grayburn, MD Institute for Health Care Research and Improvement, Baylor Research Institute, Department of Cardiothoracic Surgery, and Baylor Heart and Vascular Institute, Baylor University Medical Center, Department of Statistical Science, Southern Methodist University, Dallas, Texas; Department of Infectious Diseases, University of Louisville, Louisville, Kentucky; and Department of Global Clinical Operations, Edwards Lifesciences, Irvine, California Background. Atrial fibrillation (AF) is recognized as a common complication of open cardiac surgery, occurring in up to 65% of patients. The advancing age and increasing risk profile of patients receiving aortic valve replacement (AVR) surgery is expected to raise incidence of new-onset postoperative AF resulting in potentially higher risk of adverse outcomes. In the early postoperative course, new-onset post-avr AF is considered relatively easy to treat and is believed to have little impact on patients long-term outcome. However, the effect of new-onset post-avr AF on long-term survival is unclear. Methods. Survival was assessed in 1,039 consecutive patients without preoperative AF who underwent AVR with or without simultaneous coronary artery bypass graft at Baylor University Medical Center, Dallas, Texas between January 1, 1997 and December 31, 2006. Results. Ten-year unadjusted survival was 50.8% for patients with new-onset postoperative AF and 59.4% for patients without. A propensity-adjusted model controlling for risk factors identified by the Society of Thoracic Surgeons and other clinical-nonclinical details was used to investigate the association between new-onset AF post-avr and survival. After adjustment, new-onset AF post-avr was significantly associated with increased risk of death (hazard ratio: 1.48; 95% confidence interval 1.12 to 1.96). Conclusions. This study provides evidence that newonset post-avr AF is significantly associated with increased long-term risk of mortality independent of the preoperative severity of disease. After controlling for a comprehensive array of risk factors associated with post- AVR adverse outcomes, risk of long-term mortality in patients who developed new-onset post-avr AF was 48% higher than in patients without it. (Ann Thorac Surg 2010;90:474 80) 2010 by The Society of Thoracic Surgeons Atrial fibrillation (AF) is recognized as a common complication of open cardiac surgery, with a reported incidence of up to 65% [1 3]. Concurrent advances in surgical techniques and perioperative care allowing higher risk patients to undergo cardiac surgery, and advances in continuous monitoring technology improving detection of AF have resulted in an apparent increase in the incidence of postoperative AF [4]. While early studies allowed some to dismiss postoperative AF as benign, transient, and self-limited... and of no consequence [1], more recent research has found associations with several adverse outcomes, including increased length of stay (overall and in intensive care), risk of stroke, need for a permanent pacemaker, and more importantly in-hospital and long-term mortality [5 14]. However, much of this work, and particularly that examining long-term outcomes, has focused on outcomes after isolated coronary artery bypass graft surgery (CABG) [9, 12, 13]. For patients undergoing valve surgery or simultaneous CABG and valve surgery, much of the published research has focused on identification of risk factors for AF, including increased age, increased left atrial size, surgery for aortic regurgitation as opposed to stenosis, and concomitant CABG and valve surgery [15 18]. The question of how postoperative AF impacts outcomes of valve surgery has less definitive answers. An association between postoperative AF and increased risk of stroke in patients who had isolated aortic valve replacement (AVR) [16] or in the combined population of patient who underwent CABG, CABG and valve replacement, or heart transplantation [6] has been observed. Evidence regarding impact on long-term mortality is sparse and inconclusive. One study [16] showed postoperative AF to be an independent risk factor for early Accepted for publication Feb 26, 2010. Address correspondence to Dr Filardo, Institute for Health Care Research and Improvement, 8080 N Central Expy, Ste 500, Dallas, TX 76206; e-mail: giovanfi@baylorhealth.edu. Dr Hamilton discloses a financial relationship with Edwards Lifesciences. 2010 by The Society of Thoracic Surgeons 0003-4975/$36.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2010.02.081

Ann Thorac Surg FILARDO ET AL 2010;90:474 80 NEW-ONSET POST-AVR AF AND LONG-TERM SURVIVAL 475 postoperative mortality in patients undergoing concomitant AVR and CABG, but noted this result should be interpreted with caution based on the small sample size and low event rate. A second study showed no impact on in-patient mortality, either in the total population of patients undergoing CABG, AVR, or concomitant CABG and AVR or in any of these surgical subgroups, and noted that, while 1-year mortality was increased with postoperative AF for CABG patients, no similar effect was observed for either AVR or concomitant AVR and CABG [8]. Likewise, when 5-year survival was examined in patients who underwent aortic valve replacement or concomitant CABG and valve replacement, postoperative AF showed no significant impact [7]. However, a recent study investigating the effect of new-onset AF on long-term survival in patients receiving either isolated valvular surgery (aortic or mitral) or concomitant AVR and CABG found reduced survival associated with newonset AF though the estimated effects did not reach statistical significance [14]. To address the lack of conclusive information, we examined the relationship between new-onset postoperative AF and long-term survival in a recent and relatively large population that underwent either AVR or AVR with simultaneous CABG (AVR CABG) within the Baylor Health Care System from 1997 to 2006, considering risk factors for postoperative adverse events recognized by the Society of Thoracic Surgeons (STS) [19]. Patients and Methods Patient Data All consecutive patients who underwent either AVR or AVR CABG surgery at Baylor University Medical Center (Dallas, TX) between January 1, 1997 and December 31, 2006 without preoperative AF were considered for this study. Clinical and nonclinical details collected by Baylor University Medical Center s and the STS Adult Cardiac Surgery Database [20] were considered for the study. Data abstraction methods and procedures have been described elsewhere [20, 21]. The STS risk factors for postcardiac adverse outcomes [17, 19] are the key elements for this study. These included the following: age, gender, race, body surface area (m 2 ), diabetes, preoperative renal failure, preoperative creatinine level, chronic lung disease, hypertension, peripheral vascular disease, cerebrovascular disease, smoking status, congestive heart failure, previous revascularization procedure, myocardial infarction (MI) timing (hours since MI), preoperative angina, ejection fraction, left main disease, concomitant procedure (CABG), and urgency of operation (status). The present study obtained Institutional Review Board approval from Baylor Research Institute in May 2008. The Institutional Review Board also approved the waiver of informed consent for this project. The initial patient cohort contained a total of 1,093 subjects. Forty-two patients (3.8%) who had tricuspid Table 1. Society of Thoracic Surgeons Operative (In- Hospital) Mortality Risk Factors and Other Clinical- Nonclinical Details for 1,039 Patients Without Preoperative Atrial Fibrillation Who Underwent Aortic Valve Replacement Surgery and Survived Past the Day of Surgery at Baylor University Medical Center (Dallas, TX) Between Jan 1997 and Dec 2006 Characteristics No Postoperative AF (n 659; 63%) Postoperative AF (n 380; 37%) Demographics: Age 65.7 [54.4, 74.4] a 74.2 [66.7, 80.1] a Gender Male 63.9% 63.9% Female 36.1% 36.1% BSA 2.0 [1.8, 2.1] a 1.9 [1.8, 2.1] a Race White 80.3% 86.1% Black 9.1% 5.3% Hispanic 3.5% 0.3% Other/Unknown 7.1% 8.4% Risk factors: Diabetes mellitus 23.2% 20.0% Renal failure 5.6% 4.7% Creatinine 1.0 [0.8, 1.2] a 1.0 [0.8, 1.2] a Chronic lung disease 14.9% 21.3% Hypertension 59.5% 64.2% Peripheral vascular disease 14.4% 15.8% Cerebrovascular disease 6.4% 5.5% Endocarditis 5.9% 2.4% Smoking: Never 48.1% 52.9% Previous 33.2% 35.0% Current 18.7% 12.1% Congestive heart failure 33.5% 36.3% Previous interventions: Previous PCI 4.2% 8.4% Previous CABG 6.2% 8.7% Previous valve 4.7% 3.7% Preoperative cardiac status: MI timing None 90.3% 87.4% 6 hours 7.6% 11.1% 6 hours 2.1% 1.6% Preoperative angina 29.4% 29.0% Hemodynamics and cath: Ejection fraction (%) 50.0 [40.0, 60.0] a 50.0 [40.0, 60.0] a Left main disease 4.7% 6.8% Concomitant procedures: CABG 44.8% 59.2% Operative: Elective surgery 91.5% 90.8% a Median value (lower quartile, upper quartile). AF new-onset atrial fibrillation; BSA body surface area; CABG coronary artery bypass grafting; cath catheterization; MI myocardial infarction; PCI percutaneous coronary intervention.

476 FILARDO ET AL Ann Thorac Surg NEW-ONSET POST-AVR AF AND LONG-TERM SURVIVAL 2010;90:474 80 Table 2. Unadjusted and Propensity-Adjusted Wald 2 of the Association Between Each Potential Confounder Included in the Propensity Model and New-Onset Atrial Fibrillation for 1,039 Patients Without Preoperative Atrial Fibrillation Who Underwent Aortic Valve Replacement Surgery and Survived Past the Day of Surgery at Baylor University Medical Center (Dallas, TX) Between Jan 1997 and Dec 2006 Characteristic Unadjusted p Value p Value After Propensity Adjustment Demographics: Age 0.0001 0.8301 Gender 0.9838 0.9444 BSA 0.2143 0.9969 Race 0.0006 0.9988 Risk factors: Diabetes mellitus 0.2286 0.8495 Renal failure 0.5433 0.9900 Creatinine 0.6786 0.6711 Chronic lung disease 0.0084 0.9688 Hypertension 0.1323 0.9427 Peripheral vascular disease 0.5496 0.9766 Cerebrovascular disease 0.5820 0.9573 Endocarditis 0.0112 0.8724 Smoking 0.0224 0.9985 Congestive heart failure 0.3642 0.9124 Previous interventions: Previous PCI 0.0064 0.8706 Previous CABG 0.1389 0.9644 Previous Valve 0.4378 0.9540 Preoperative cardiac status: MI timing 0.1473 0.9958 Preoperative angina 0.8669 0.9524 Hemodynamics and cath: Ejection fraction 0.3664 0.9093 Left main disease 0.1472 0.9397 Concomitant procedures: CABG 0.0001 0.9815 Operative: Elective surgery 0.7998 0.8503 BSA body surface area; CABG coronary artery bypass grafting; cath catheterization; MI myocardial infarction; PCI percutaneous coronary intervention. surgery, preoperative use of an intraaortic balloon pump, or missing MI timing were not considered. Likewise, patients who died on day of surgery (n 12; 1.1%) were not included in the study. The final patient cohort included 1,039 (95.1%). Exposure and Outcome Definition New-onset postoperative AF was defined as AF (as classified by the STS Adult Cardiac Surgery Database) occurring during hospitalization after AVR or AVR CABG in a patient with no history of AF. Accordingly, a patient without preoperative AF who, according to a clinician (either a doctor or a nurse) experienced a sweep of a tele strip at any time during hospitalization after surgery, indicating AF lasting more than 10 minutes that required medical treatment was classified as a patient with newonset post-avr AF (independent of whether the patient responded to medical-surgical management or spontaneously reverted to sinus rhythm). Survival was measured as time (in days) to either death or last follow-up from date of surgery. Vital status was assessed in all 1,039 patients using the National Death Index data from January 1, 1997 to November 1, 2007; the completeness of the follow-up was 100%. Statistical Analysis A Cox proportional hazards model (without covariates) was employed to estimate the unadjusted hazard ratio of the association between new-onset post-avr AF and long-term mortality, while the unadjusted survival curves and accompanying survival estimates were estimated using the Kaplan and Meier method. Risk factors by which these survival curves were adjusted for are listed in Table 1. Differences in these factors between the new-onset AF and no AF groups were tested by a Wilcoxon-Mann-Whitney test (continuous factors) or a 2 test (categoric factors). To address the issue of statistical multiplicity, the p values were adjusted with the method of Bonferroni (differences that are significant at alpha 0.05 are reported in the results section). To provide an adjusted test of the association between new-onset postoperative AF and long-term survival, a propensity score approach was used [22, 23]. The propensity score was developed by fitting a logistic regression model with new-onset postoperative AF as the outcome variable and all factors in Table 1 as the predictor variables. Missing data in this model were accounted for through multiple imputation using predictive mean matching [22, 24] (0.2% missing status, 21.8% missing ejection fraction, 37.6% missing creatinine, 2.4% missing BSA). To check that balance was achieved with the propensity model between the new-onset AF and no new-onset AF groups we statistically compared balance between these groups for the risk factors listed in Table 1 both before and after adjustment by the propensity score. Specifically, we estimated unadjusted (logistic regression) Wald 2 for the association of each preoperative risk factor and new-onset AF. We then adjusted each association by the propensity as a covariate. The Wald 2 estimated from these adjusted associations were compared with the unadjusted 2 to check for balance; all the 2 p values after adjustment for the propensity score were not statistically significant (see Table 2) indicating very good balance after adjustment. The resulting propensity scores from this model were then used to adjust the effect of new-onset postoperative AF

Ann Thorac Surg FILARDO ET AL 2010;90:474 80 NEW-ONSET POST-AVR AF AND LONG-TERM SURVIVAL 477 on long-term mortality. Specifically, if we let y p 1 denote the linear predictor predicted value (X ) from the propensity model described above and Afib denote whether the patient experienced AF, then the final model was Here log(h(t)) 0 Afib 1 y p (1) 4 y p (4). y p (j 1) a j a 4 (k 5 k j ) (k 5 k 4 ) a 5 (k 4 k j ) (k 5 k 4 ) for j 1 to 3, where k j denotes the j th knot for the cubic spline and a j y p 1 k j 3 if y p 1 k j and 0 otherwise. Possible effect modification produced by age, gender, and concomitant CABG was investigated; statistical tests for interactions were not significant. Likewise the proportionality of the hazards in the above cubic spline proportional hazard model was tested by the Grambsch and Therneau method [25]. The adjusted survival curves were created by the method of Kalbfleisch and Prentice [26]. All analyses were performed using R version 2.8.1. Results Thirty-seven percent (n 380) of patients experienced new-onset postoperative AF. Patients experiencing newonset postoperative AF were in general older (p value 0.001), were more likely to be undergoing CABG (p value 0.001), and were more likely to be white (p value 0.016) compared to patients who did not experience new-onset AF (Table 1). The total follow up years were 4,551.3. The unadjusted survival analysis showed a significant difference between those patients experiencing new onset postoperative AF Fig 2. Adjusted (proportional hazard model) curves and 95% confidence intervals depicting the effect of new-onset postoperative AF on survival in patients who underwent aortic valve replacement surgery and survived past the day of surgery at Baylor University Medical Center (Dallas, TX) United States between January 1997 and December 2006. The propensity adjusted model includes the following risk factors: age, gender, body surface area, race, diabetes, preoperative renal failure, preoperative creatinine level, chronic lung disease, hypertension, peripheral vascular disease, cerebrovascular disease, endocarditis, smoking status, congestive heart failure, previous revascularization procedure (percutaneous coronary intervention or CABG), previous valve surgery, myocardial infarction timing (time since myocardial infarction), ejection fraction, left main disease, urgency of operation (elective or nonelective), and concomitant procedure (CABG). For the purposes of the adjusted curves, all factors are set to their mean values. (Afib patient with new-onset postoperative AF; No Afib Patient without new-onset postoperative atrial fibrillation; AF atrial fibrillation; CABG coronary artery bypass grafting.) Fig 1. Unadjusted (Kaplan-Meier) curves depicting the effect of new-onset postoperative atrial fibrillation on survival in patients who underwent aortic valve replacement surgery and survived past the day of surgery at Baylor University Medical Center (Dallas, TX) United States between January 1997 and December 2006. (Afib patient with new-onset postoperative atrial fibrillation; No Afib patient without new-onset postoperative atrial fibrillation.) and those who did not (p value 0.001) (Fig 1). The unadjusted hazard ratio was 1.68 (95% CI: 1.30 to 2.16). The unadjusted 5-year survival estimates for patients experiencing new-onset postoperative AF was 0.708 (95% CI: 0.651 to 0.758) as opposed to 0.831 (95% CI: 0.797 to 0.860) for those who did not. The difference was similar at 10 years: the survival estimates were 0.508 (95% CI: 0.404 to 0.603) and 0.594 (95% CI: 0.485 to 0.687) for the AF and no AF groups, respectively. After adjustment (via the propensity score model) for the factors in Table 1, the difference in the survival curves for the two patient groups remained highly significant (p value 0.005). The adjusted hazard ratio was 1.48 (95% CI: 1.12 to 1.96), indicating that patients who experienced new-onset postoperative AF were at a 48% higher risk of death than patients who did not. The adjusted survival curves in Figure 2 illustrate the difference between the two groups. Comment Our study results complement the existing data regarding the harmful effect of new-onset postoperative AF on

478 FILARDO ET AL Ann Thorac Surg NEW-ONSET POST-AVR AF AND LONG-TERM SURVIVAL 2010;90:474 80 long-term survival after CABG [12 14] and extend it with evidence concerning aortic valve surgery. Mariscalco and Engström [14] previously observed a nonsignificant negative effect of new-onset AF after valvular surgery; our significant negative results provide compelling evidence of this association. Our data showed that, in patients undergoing AVR, new-onset postoperative AF is strongly associated with increased long-term risk of mortality independent of patients preoperative severity (as defined by the combination of available preoperative data). Risk of long-term mortality in patients who developed new-onset post- AVR AF was significantly higher (48%) than in patients without it, even after controlling for a comprehensive array of risk factors and confounders associated with post-avr adverse outcomes. These results indicate that a patient in this cohort who developed new-onset postoperative AF had a 48% higher risk of mortality than a patient with exactly the same preoperative risk profile that did not. We do not know why patients that experienced newonset AF after undergoing aortic valve surgery suffered reduced survival. We hypothesize that some of these patients would develop chronic or recurrent AF and therefore would ultimately experience higher risk of stroke and congestive heart failure years after surgery. Unfortunately, we do not have the postdischarge AF data that would shed further light on the impact of progression of new-onset AF on patient survival. Study limitations that should be considered in interpreting our results include the following: (1) generalizability might be limited given that the study was conducted at a single center in Dallas, Texas; (2) new-onset postoperative AF may be a marker of underlying myocardial disease, inflammation, or neurohormonal activity associated with poorer survival [27]; (3) the lack of available data on the preoperative, intraoperative, and postoperative management (including pharmacologic intervention) of AF, and of clinical data regarding the etiology of the valve disease, which meant their confounding effects were not accounted for in our study (however, the multivariable propensity adjusted analysis accounted for an extensive list of risk factors identified by the STS, in addition to other important variables providing a rigorous adjustment for potential confounders); and (4) as in any observational study, causality between new-onset AF and mortality could not be definitively established and further research is necessary to assess it (however, a randomized clinical trial to assess causality is not an option as we clearly cannot randomize patients to newonset AF or no new-onset AF). Our findings nonetheless have important public health implications as the advancing age [15] and increasing risk profile of patients receiving AVR surgery is expected to raise incidence of new-onset postoperative AF, resulting in potentially poorer long-term survival. Current guidelines reflect the notion that new-onset postoperative AF is a minor and transient short-term adverse outcome, and that does not significantly affect long-term outcomes such survival [12, 28, 29]. The American College of Cardiology/American Heart Association Task Force and the European Society of Cardiology (ACC/AHA/ESC) clinical guidelines [30] do not specifically address primary prevention, focusing rather on the management of new-onset postoperative. Nevertheless, the American College of Chest Physicians (ACCP) guidelines for the prevention and management of postoperative AF after cardiac surgery [31] stress that more research is needed in the area of the prevention and management of postoperative AF. Our data provide evidence regarding the long-term consequences associated with new-onset post- AVR AF and pose important questions regarding whether compliance with current guidelines for management of postoperative AF are effective. Unfortunately, available registries (eg, STS database) do not collect detailed information on postoperative management, making the assessment of optimal management (based on AHA/ACC/ESC and AACP guidelines) of postoperative AF and its impact on survival problematic. Likewise, data on optimal prophylaxis rates and more importantly on its effectiveness in preventing new-onset postoperative AF are lacking. Accordingly, to reduce the incidence of new-onset postoperative AF and ultimately improve survival in patients undergoing AVR surgery, future research should focus on the following: (1) investigating rates of optimal prophylaxis for and management of postoperative AF; (2) assessing the effectiveness of optimal prophylaxis for and management of postoperative AF; (3) developing and implementing initiatives aimed at increasing compliance with established preventative and postoperative management strategies; (4) rigorously assessing preoperative, intraoperative, and postoperative risk factors for new onset postoperative AF; and (5) developing new strategies for prophylaxis and management of postoperative AF. We would like to acknowledge the use of software from Professor Frank Harrell s Hmisc and Design libraries, and to thank Briget da Graca for writing and editorial assistance. Grant support was provided by the Cardiovascular Research Review Committee in cooperation with the Baylor Heart and Vascular Institute. References 1. Almassi GH, Schowalter T, Nicolosi AC, et al. Atrial fibrillation after cardiac surgery: a major morbid event? Ann Surg 1997;226:501 513. 2. Maisel WH, Rawn JD, Stevenson WG. Atrial fibrillation after cardiac surgery. Ann Intern Med 2001;135:1061 73. 3. Shantsila E, Watson T, Lip GY. Atrial fibrillation postcardiac surgery: changing perspectives. Curr Med Res Opin 2006;22:1437 41. 4. 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 7. 5. Chaput M, Bouchard D, Demers P, et al. Conversion to sinus rhythm does not improve long-term survival after valve surgery: insights from a 20-year follow-up study. Eur J Cardiothorac Surg 2005;28:206 10.

Ann Thorac Surg FILARDO ET AL 2010;90:474 80 NEW-ONSET POST-AVR AF AND LONG-TERM SURVIVAL 479 6. Creswell LL, Schuessler RB, Rosenbloom M, Cox JL. Hazards of postoperative atrial arrhythmias. Ann Thorac Surg 1993; 56:539 49. 7. Mahoney EM, Thompson TD, Veledar E, Williams J, Weintraub WS. Cost-effectiveness of targeting patients undergoing cardiac surgery for therapy with intravenous amiodarone to prevent atrial fibrillation. J Am Coll Cardiol 2002;40: 737 45. 8. Mariscalco G, Engström KG. Atrial fibrillation after cardiac surgery: Risk factors and their temporal relationship in prophylactic drug strategy decision. Int J Cardiol 2007;129: 354 62. 9. Mariscalco G, Klersy C, Zanobini M, et al. Atrial fibrillation after isolated coronary surgery affects late survival. Circulation 2008;118:1612 8. 10. Ruel M, Kulik A, Lam BK, et al. Long-term outcomes of valve replacement with modern prostheses in young adults. Eur J Cardiothorac Surg 2005;27:425 33. 11. Ruel M, Masters RG, Rubens FD, et al. Late incidence and determinants of stroke after aortic and mitral valve replacement. Ann Thorac Surg 2004;78:77 84. 12. Villareal RP, Hariharan R, Liu BC, et al. Postoperative atrial fibrillation and mortality after coronary artery bypass surgery. J Am Coll Cardiol 2004;43:742 8. 13. Filardo G, Hamilton C, Hebeler RF Jr, Hamman B, Grayburn P. New-onset post-operative atrial fibrillation following isolated coronary artery bypass graft surgery and long-term survival. Circulation: Cardiovascular Quality and Outcomes 2009;2:164 9. 14. Mariscalco G, Engström KG. Postoperative atrial fibrillation is associated with late mortality after coronary surgery, but not after valvular surgery. Ann Thorac Surg 2009;88:1871 6. 15. Census. US Census Report: We the People, Aging in the United States. Available at: http://www.census.gov/population/ www/socdemo/age/general-age.html#older: Accessed November 24, 2009. 16. Banach M, Goch A, Misztal M, Rysz J, Jaszewski R, Goch JH. Predictors of paroxysmal atrial fibrillation in patients undergoing aortic valve replacement. J Thorac Cardiovasc Surg 2007;134:1569 76. 17. Ducceschi V, D Andrea A, Galderisi M, et al. Risk predictors of paroxysmal atrial fibrillation following aortic valve replacement. Ital Heart J 2001;2:507 12. 18. Orlowska-Baranowska E, Baranowski R, Michalek P, Hoffman P, Rywik T, Rawczylska-Englert I. Prediction of paroxysmal atrial fibrillation after aortic valve replacement in patients with aortic stenosis: identification of potential risk factors. J Heart Valve Dis 2003;12:136 41. 19. O Brien SM, Shahian DM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 2 isolated valve surgery. Ann Thorac Surg 2009;88(1 suppl): S23 42. 20. Ferguson TB Jr, Dziuban SW Jr, Edwards FH, et al. The STS National Database: current changes and challenges for the new millennium. Committee to Establish a National Database in Cardiothoracic Surgery, The Society of Thoracic Surgeons. Ann Thorac Surg 2000;69:680 91. 21. Filardo G, Hamilton C, Hamman B, Ng HK, Grayburn P. Categorizing BMI may lead to biased results in studies investigating in-hospital mortality after isolated CABG. J Clin Epidemiol 2007;60:1132 9. 22. Harrell FE Jr. Regression modeling strategies: with application to linear models, logistic regression, and survival analysis. New York: Springer-Verlag; 2001. 23. D Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265 81. 24. Shahian DM, Blackstone EH, Edwards FH, et al. Cardiac surgery risk models: a position article. Ann Thorac Surg 2004;78:1868 77. 25. Grambsch PM, Therneau T. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994; 81:515 26. 26. Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. 2nd ed: Wiley; 2002. 27. Cain ME, Curtis AB. Rhythm control in atrial fibrillation one setback after another. N Engl J Med 2008;358: 2725 7. 28. Rubin DA, Nieminski KE, Reed GE, Herman MV. Predictors, prevention, and long-term prognosis of atrial fibrillation after coronary artery bypass graft operations. J Thorac Cardiovasc Surg 1987;94:331 5. 29. Landymore RW, Howell F. Recurrent atrial arrhythmias following treatment for postoperative atrial fibrillation after coronary bypass operations. Eur J Cardiothorac Surg 1991;5: 436 9. 30. Fuster V, Ryden LE, Cannom DS, et al. ACC/AHA/ESC 2006 Guidelines for the Management of Patients with Atrial Fibrillation: a report of the American College of Cardiology/ American Heart Association Task Force on Practice Guidelines and the European Society of Cardiology Committee for Practice Guidelines (Writing Committee to Revise the 2001 Guidelines for the Management of Patients With Atrial Fibrillation): developed in collaboration with the European Heart Rhythm Association and the Heart Rhythm Society. Circulation 2006;114:e257 354. 31. American College of Chest Physicians guidelines for the prevention and management of postoperative atrial fibrillation after cardiac surgery. Chest 2005;128(1 suppl):61s 4S. INVITED COMMENTARY Filardo and colleagues [1] report that early postoperative atrial fibrillation (AF) is associated with a significant reduction in late survival for patients undergoing aortic valve replacement, with or without concomitant coronary artery bypass grafting or mitral valve repair or replacement. The authors suggest that increased efforts to prevent this postoperative complication, and the use of better monitoring of therapeutic interventions are therefore warranted. We know from longitudinal studies in the general population that AF is associated with a variety of morbidities, as well as a reduction in long-term survival. Unfortunately, there is little information regarding the long-term course of postcardiac patients who have early postoperative AF develop. In one recent report involving a much larger group of patients, Mariscalco and Engstrom [2] reported that there is reduced long-term survival for coronary artery bypass grafting patients who have early postoperative AF develop. In contrast to the current report, however, Mariscalco and Engstrom [2] did not find any such relationship for patients who underwent valve surgery. It is not clear from these two reports why the findings might be so different for valve surgery patients, including those who underwent aortic valve replacement. These findings are important because they suggest that the complication of postoperative AF may be more significant than we often believe. It may well be the case that patients with this complication, even if treated success- 2010 by The Society of Thoracic Surgeons 0003-4975/$36.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2010.03.033