- Email: [email protected]
Surgical Ablation of Atrial Fibrillation in the United States: Trends and Propensity Matched Outcomes
CARDIOTHORACIC ANESTHESIOLOGY: The Annals of Thoracic Surgery CME Program is located online at http://www.annalsthoracicsurgery. org/cme/home. To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal.
Vinay Badhwar, MD, J. Scott Rankin, MD, Niv Ad, MD, Maria Grau-Sepulveda, MD, MPH, Ralph J. Damiano, MD, A. Marc Gillinov, MD, Patrick M. McCarthy, MD, Vinod H. Thourani, MD, Rakesh M. Suri, MD, DPhil, Jeffrey P. Jacobs, MD, and James L. Cox, MD Department of Cardiovascular and Thoracic Surgery, West Virginia University, Morgantown, West Virginia; Duke Clinical Research Institute, Durham, North Carolina; Division of Cardiothoracic Surgery, Washington University, St. Louis, Missouri; Department of Thoracic and Cardiovascular Surgery, Cleveland Clinic, Cleveland, Ohio; Division of Cardiac Surgery, Northwestern University Feinberg School of Medicine, Chicago, Illinois; Division of Cardiothoracic Surgery, Emory University, Atlanta, Georgia; and Division of Cardiac Surgery, Johns Hopkins University School of Medicine, Baltimore, Maryland
Background. Surgical ablation (SA) for atrial fibrillation (AF) effectively restores sinus rhythm. Incompletely defined risk has previously limited concomitant performance of SA during cardiac operations. The study goals were to define performance trends and risk-adjusted outcomes for contemporary SA. Methods. From July 2011 to June 2014, 86,941 patients with AF, but without endocarditis, underwent primary nonemergent cardiac operations in The Society of Thoracic Surgeons (STS) database. Cochran-Armitage tests examined performance trends of SA for six operative categories: mitral valve repair or replacement (MVRR) with or without coronary artery bypass graft surgery (CABG), aortic valve replacement (AVR) with or without CABG, CABG, AVR with MVRR, stand-alone SA, and other concomitant operations. The risk of concomitant SA was analyzed by propensity matching 28,739 patient-pairs with and without SA by AF type, primary operation, and STS comorbid risk variables using greedy 1:1 matching algorithms. Results. Among all patients with AF, 48.3% (42,066 of 86,941) underwent SA. Mitral operations had the highest
trial fibrillation (AF) at the time of cardiac operations negatively affects 30-day outcome and survival [1–5]. Surgical ablation (SA) restores normal sinus rhythm and improves long-term quality of life [6–10]. Prior registry trends revealed that adoption of SA concomitant to cardiac operations was limited, and concerns over added
A
rate of SA (MVRR ± CABG 68.4% [14,693 of 21,496]; MVRR D AVR 59.1% [1,626 of 2,750]). The AVR ± CABG and isolated CABG rates were 39.3% (6,816 of 17,349) and 32.8% (9,156 of 27,924), respectively. Nearly half of other concomitant operations underwent SA, 47.6% (6,939 of 14,586). Performance frequency increased throughout the study period. After propensity matching, SA was associated with a reduction in relative risk (RR) of 30-day mortality (RR 0.92, 95% confidence interval [CI]: 0.85 to 0.99) and stroke (RR 0.84, 95% CI: 0.74 to 0.94), but an increase in renal failure (RR 1.12, 95% CI: 1.03 to 1.22) and pacemaker implantation (RR 1.33, 95% CI: 1.24 to 1.43). Conclusions. Contemporary utilization of SA is increasing across all operative categories. Performance of SA is accompanied by a 30-day reduction in mortality and stroke. These findings further refine our understanding of the role of SA in the treatment of AF. (Ann Thorac Surg 2017;104:493–500) Ó 2017 by The Society of Thoracic Surgeons
Dr Rankin discloses a financial relationship with AtriCure and Medtronic; Dr Ad with Medtronic, Atricure, LivaNova, Nido Surgical, and Left Atrial Appendage Occlusion LLC; Dr Gillinov with AtriCure, Medtronic, Abbott, St. Jude, Edwards, and Clearflow; Drs Damiano and Cox with AtriCure.
Accepted for publication May 5, 2017. Presented at the Sixty-third Annual Meeting of the Southern Thoracic Surgical Association, Naples, FL, Nov 9–12, 2016. Address correspondence to Dr Badhwar, Department of Cardiovascular and Thoracic Surgery, West Virginia University, 1 Medical Center Dr, Morgantown, WV 26506-8059; email: [email protected].
Ó 2017 by The Society of Thoracic Surgeons Published by Elsevier Inc.
The Supplemental Table and Figure can be viewed in the online version of this article [http://dx.doi. org/10.1016/j.athoracsur.2017.05.016] on http://www. annalsthoracicsurgery.org.
0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2017.05.016
ADULT CARDIAC
Surgical Ablation of Atrial Fibrillation in the United States: Trends and Propensity Matched Outcomes
494
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
ADULT CARDIAC
Abbreviations and Acronyms ACSD AF AVR CABG CI LA MVRR RF RR SA STS
= = = = = = = = = = =
Adult Cardiac Surgery Database atrial fibrillation aortic valve replacement coronary artery bypass graft surgery confidence interval left atrium mitral valve repair or replacement radiofrequency relative risk surgical ablation The Society of Thoracic Surgeons
operative risk may have influenced clinical decision making [11]. Although techniques for SA and experience have improved in recent years, contemporary trends of SA remain incompletely defined. Moreover, for AF patients, recent information suggests operative mortality after cardiac surgery may be improved with SA [12]. The goal of this study was to define performance trends and comparatively assess outcomes of contemporary SA using The Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database (ACSD).
Material and Methods Patient Populations From July 1, 2011, to June 30, 2014, 837,978 cardiac operations were recorded in the STS ACSD; of those, 112,401 (13.4%) had preoperative AF documented. Exclusions were emergent status, endocarditis, reoperation, mechanical circulatory support, transplant, aortic dissection, and pulmonary thromboembolectomy. After exclusions, three populations were studied to examine three objectives: (1) all patients receiving SA were examined for technical procedural trends of SA over time, regardless of AF type documentation; (2) all patients with documented AF type were examined to descriptively assess patients treated with SA or not treated with SA (no-SA); and (3) the risk of adding concomitant SA to a primary cardiac operation was assessed by comprehensive propensity matched comparisons inclusive of AF type, primary operation, and comorbidities, after excluding stand-alone operations owing to lack of a no-SA comparator (Fig 1). In a first trends analysis, all 62,025 SA patients were utilized to document overall performance trends and energy source utilization over time in six operative categories: isolated coronary artery bypass graft surgery (CABG), mitral valve repair or replacement (MVRR) with or without CABG, aortic valve replacement (AVR) with or without CABG, AVR with or without MVRR, stand-alone SA, and other concomitant operations. A second comparative descriptive analysis of SA versus no-SA included all 86,941 patients with documented AF type, after operative exclusions. These were assessed in total and by each of the six operative categories. Finally, a third comparative outcomes analysis assessed the risk of
Ann Thorac Surg 2017;104:493–500
adding SA as a concomitant procedure. This analysis included all the above concomitant operations by excluding the 2,836 stand-alone SA patients. The resulting cohort of 84,105 patients with documented AF was evaluated by a propensity matched analysis of concomitant SA versus no-SA.
Baseline Characteristics All relevant procedural STS covariates and baseline characteristics were evaluated using medians with 25th and 75th percentiles for continuous variables and proportions for categoric variables. Descriptive data were compared between groups using the Wilcoxon rank sum test for continuous and ordinal variables, and a Pearson c2 test or Fisher’s exact test for categoric variables, as appropriate.
Trends Analysis Temporal trends in SA utilization across the six operative categories were evaluated with Cochran-Armitage trends tests. The 3 years of data were divided into six 6-month sextiles: period 1 (July 1, 2011, to December 31, 2011) to period 6 (January 1, 2014, to June 30, 2014). Total numbers of procedures and incidence of SA for each procedural type were assessed.
Propensity-Matched Comparative Outcome Analysis To quantify risk-adjusted outcome differences for SA versus no-SA with documented preoperative AF type, a comprehensive propensity matched analysis was performed. Using multivariate logistic regression, probability of treatment assignment to SA versus no-SA (propensity score) was generated from observed covariates and procedures. Missing values (less than 3%) underwent simple imputation in accordance with validated STS models [12]. Group-specific medians were examined for continuous variables, and most common category for categoric variables. Continuous variables were evaluated using restricted cubic spline plots to assess linearity. The final propensity model included the following: AF type (paroxysmal versus persistent), age (linear splines with knot at 75 years), sex, body surface area, creatinine (linear splines with knot at 1.00), ejection fraction (linear splines with knot at 50%), body surface area (quadratic spline), multiplicative interaction between body surface area and sex, New York Heart Association class III to IV, cerebrovascular disease, stroke, chronic lung disease, peripheral vascular disease, home oxygen, sleep apnea, syncope, diabetes type, preoperative myocardial infarction, operative status, previous cardiovascular interventions, previous pacemaker, mitral and tricuspid insufficiency, and operative procedure type (CABG, MVRR CABG, AVR CABG, AVR þ MVRR, other concomitant operations). Propensity scores matched SA and no-SA using a 1:1 greedy 5-to-1 digitmatching algorithm. Only pairs matched on three or more propensity score digits were retained in the final matched sample of SA (n ¼ 28,739) and no-SA (n ¼ 28,739; Supplemental Table). Propensity score distribution and protocol adequacy were assessed to a uniform
Ann Thorac Surg 2017;104:493–500
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
495
1% standardized difference or less between matched pairs (Supplemental Figure). Outcome risk ratios and 95% confidence intervals (CI) for SA versus no-SA were generated for unadjusted and propensity matched comparisons, using generalized estimating equation–modified Poisson regression models with a log-link and an independent correlation structure to account for hospital or participant site clustering. All analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC). A p value less than 0.05 was considered significant. This study of deidentified data from the STS ACSD has been granted a waiver of informed consent by the Duke University Institutional Review Board.
Results Procedural Trends of Surgical Ablation Overall volume of SA showed incremental growth progressively through each of the 6-month sextiles as well as an overall increase of 50% during the study period (p < 0.0001). When analyzed by operation category, significant upward trends in SA rates were observed for all groups except the two least frequent operations (AVR þ MVRR and stand alone; Fig 2). Procedural details are outlined in Table 1. Among all 62,025 SA operations, the ablation location was recorded as left atrium (LA) only in 52.9%, biatrial in 35.8%, 1.7% right atrium only, and 9.6% missing. The left atrial appendage was surgically obliterated in 86.4% overall. Radiofrequency (RF) ablation was used in 49.2% of all
cases, cryoablation in 27.3%, and cut-and-sew lesions applied in 13.1%. Only 2.5% of all cases had ultrasound, microwave, or laser energy applied. Energy source type was missing in 7.9%. Examining specific ablation strategies applied, RF only was used in 36.5%, cryoablation only in 15.3%, cut-and-sew lesions only in 8.4%, and RF plus cryoablation combination in 9.2%, with other combinations being rare.
Contemporary Performance of Surgical Ablation for Atrial Fibrillation In patients with documented AF (n ¼ 86,941), the overall incidence of SA was 48.3%. In this cohort, the most common operation performed was isolated CABG (33.1%), but that had the lowest rate of concomitant SA (33.0%). A MVRR with or without CABG was performed in 25.0%, but had the highest rate of SA (68.4%). The frequency of concomitant SA in AVR with MVRR was 59.1%, other concomitant procedures was 47.6%, and AVR with or without CABG was 39.3%. Baseline characteristics were generally better in SA patients, but the incidences of severe mitral regurgitation and permanent AF were higher (Table 2).
Comparative Outcome Analysis With or Without Concomitant Surgical Ablation Unadjusted operative mortality and most postoperative morbidities were lower with SA, except for permanent pacemaker rates, which were higher in SA versus no-SA (7.6% versus 4.4%, respectively, p < 0.0001). Readmission rates within 30 days were similar (Table 3).
ADULT CARDIAC
Fig 1. Flow diagram of study populations using Consolidated Standards of Reporting Trials (CONSORT) principles. Source: The Society of Thoracic Surgeons Adult Cardiac Database, July 1, 2011, through June 30, 2014. (AF ¼ atrial fibrillation.)
ADULT CARDIAC
496
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
Ann Thorac Surg 2017;104:493–500
Fig 2. Trends in total numbers of surgical ablation procedures in the United States, July 1, 2011, through June 30, 2014. (AVR ¼ aortic valve replacement; CABG ¼ coronary artery bypass graft surgery; MVRR ¼ mitral valve repair or replacement.)
Analyzing 28,739 propensity matched pairs revealed that SA had lower risk-adjusted mortality (RR 0.92, 95% CI: 0.85 to 0.99), permanent stroke (RR 0.84, 95% CI: 0.74 to 0.94), and prolonged ventilation (RR 0.95, 95% CI: 0.90 to 0.99). Pacemaker implantation was more prevalent in the SA group (RR 1.33, 95% CI: 1.24 to 1.43), as was postoperative renal failure (RR 1.12, 95% CI: 1.03 to 1.22) and overall 30-day readmissions (RR 1.09, 95% CI: 1.03 to 1.15). Phrenic nerve injury remained very rare (0.06%) and was equal between groups (Table 4).
Comment This study illustrates the rising utilization of SA concomitant to cardiac operations in the United States. Importantly, by examining 57,478 propensity-matched
patients, the performance of SA was protective of stroke, prolonged ventilation, and 30-day operative mortality, but was associated with an increase in pacemaker risk and perioperative renal failure. The benefit of SA is to restore normal sinus rhythm and quality of life, especially when performed concomitant to mitral surgery [6, 9–11, 13–16]. Nevertheless, registry investigations at earlier sampling periods did not document widespread adoption [11, 17]. A 2005 to 2010 STS ACSD series noted slight decreasing trends over time, with only 40.6% of AF patients undergoing SA [17]. Within the past 5 years, clinical interest in standardized SA techniques have grown substantially [18–20]. These correlate with the Food and Drug Administration– approved indication of SA for AF with one device manufacturer. At mitral operation, the exposed LA
Table 1. Surgical Ablation Procedures Performed by Operation Type
Variable Left atrium only location Biatrial location Atrial location not documented Primarily endocardial Primarily epicardial Ablation location not documented Radiofrequency only Cryoablation only Cut-and-sew lesions only Radiofrequency and cryoablation Left atrial appendage obliterated
Isolated Other Stand MVRR CABG AVR CABG AVR þ MVRR CABG Concomitant Alone (n ¼ 21,992) (n ¼ 9,875) (n ¼ 2,304) (n ¼ 14,334) (n ¼ 10,252) (n ¼ 3,268) 50.9 40.0 9.1 52.4 31.2 16.4
57.9 30.7 11.4 26.6 56.3 17.1
52.1 37.9 10.0 49.3 35.6 15.1
58.0 29.3 12.7 22.0 55.8 22.2
50.5 38.1 11.4 39.9 43.4 16.7
37.2 56.3 6.5 23.6 72.5 3.9
<0.0001
27.0 23.8 7.6 11.8 87.7
43.1 8.9 8.6 7.0 88.5
28.8 19.5 9.0 11.8 86.3
42.5 7.4 10.2 5.5 88.6
34.0 15.1 9.5 9.7 85.9
67.8 9.2 0.8 11.0 63.9
<0.0001
Values are percentages. Surgical ablation techniques performed between July 1, 2011, and June 30, 2014. AVR ¼ atrial valve replacement;
p Value
CABG ¼ coronary artery bypass graft surgery;
MVRR ¼ mitral valve repair or replacement.
<0.0001
<0.0001
Ann Thorac Surg 2017;104:493–500
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
497
Table 2. Patient Characteristics in the Overall Atrial Fibrillation Population Variable
No Ablation (n ¼ 44,875)
Ablation (n ¼ 42,066)
p Value
72 (65–79) 35.0 55 (43–60) 18.2 2.8 15.0 13.4 9.4 16.6 32.0 47.9 50.7 25.2
73 (65–80) 33.6 55 (40–60) 19.6 3.9 17.1 16.1 11.0 23.0 33.4 41.6 58.4 16.9
71 (64–78) 36.7 55 (45–60) 16.7 1.7 12.7 10.6 7.7 9.8 30.5 54.6 45.4 33.9
<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001
Age, years Female Ejection fraction Cerebrovascular disease Preoperative dialysis Moderate-severe COPD Peripheral vascular disease Preoperative pacemaker Preoperative MI within 21 days NYHA class III or IV Persistent atrial fibrillation Paroxysmal atrial fibrillation Severe mitral regurgitation
Values are median (interquartile range) or percentage. Unadjusted baseline patient characteristics in patients with atrial fibrillation between July 1, 2011 and June 30, 2014. COPD ¼ chronic obstructive pulmonary disease;
MI ¼ myocardial infarction;
provides an opportune time to perform SA. In the current study, SA was performed in 48.3% of all patients with AF, and in 68% of concomitant MVRR operations—the highest among all operations. Logically, MVRR facilitated the highest frequency of primarily endocardial lesions (Table 1). Conversely, at the time of closed atrial operations such as isolated CABG or AVR with or without CABG, epicardial, RF only, and LA only lesions were most commonly applied. Many would advocate left atrial appendage obliteration for its impact on long-term morbidity reduction and need for prolonged anticoagulation. Although left atrial appendage obliteration was performed in more than 85% of concomitant SA procedures, it was only completed in 63.9% of stand-alone cases, possibly
NYHA ¼ New York Heart Association.
reflective of the recent resurgence of alternative hybrid procedures. This finding may represent an opportunity for improvement. The longitudinal merits of the full biatrial Cox maze IV lesion set, particularly for persistent AF, are known [14, 18, 19]. However, controversy persists, with recent reasonable results of LA-only lesions [6, 21]. Current practice appears to be reflective of this controversy. The most common lesions applied in all concomitant categories was LA only (Table 2). The exception was with stand-alone SA, in which primarily endocardial biatrial lesions prevail. Although these data do not document long-term outcomes, they do highlight an opportunity for further clarity by identifying which patient subsets would benefit most from LA versus biatrial lesions.
Table 3. Unadjusted Outcomes of Atrial Fibrillation Patients With or Without Concomitant Surgical Ablation Variable Operative mortality Reoperation for bleeding Permanent stroke Transient ischemic attack Prolonged ventilation >48 hours Sternal infection Permanent pacemaker Phrenic nerve injury Discharged home Discharged extended care Discharge coumadin Discharge thrombin inhibitor Discharge antiarrhythmic drug Less than 30-day readmission Pacemaker readmission
Overall (n ¼ 84,105)
No Ablation (n ¼ 44,875)
Ablation (n ¼ 39,230)
p Value
4.4 3.6 2.0 0.4 16.5 0.3 5.9 0.5 64.4 28.9 57.8 5.2 53.2 13.4 2.5
4.9 3.6 2.2 0.4 17.7 0.3 4.4 0.4 61.1 31.5 52.7 4.8 50.2 13.4 2.2
3.8 3.6 1.7 0.3 15.1 0.2 7.6 0.6 68.1 25.9 63.6 5.7 56.5 13.4 2.8
<0.0001 0.8736 <0.0001 0.0142 <0.0001 0.1398 <0.0001 0.2287 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.4497 <0.0001
Values are percentages. Concomitant surgical ablation with known atrial fibrillation performed between July 1, 2011, and June 30, 2014, after excluding 2,836 patients with stand-alone therapy.
ADULT CARDIAC
Overall (n ¼ 86,941)
498
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
Ann Thorac Surg 2017;104:493–500
Table 4. Relative Risks of Performing Concomitant Surgical Ablation in Propensity Matched Patients With Atrial Fibrillation During Adult Cardiac Operations Overall (n ¼ 57,478)
ADULT CARDIAC
Outcome Mortality Reoperation for bleeding Permanent stroke Transient ischemic attack Prolonged ventilation >48 hours Renal failure Pacemaker Phrenic nerve injury Readmission 30 days
4.31 3.61 1.96 0.38 16.31 4.62 6.87 0.06 13.36
(2,480) (2,075) (1,124) (218) (9,373) (2,585) (3,946) (33) (7,347)
No Ablation (n ¼ 28,739) 4.50 3.73 2.13 0.42 16.75 4.35 5.89 0.06 12.79
(1,292) (1,073) (612) (121) (4,813) (1,219) (1,693) (16) (3,511)
Ablation (n ¼ 28,739) 4.13 3.49 1.78 0.34 15.87 4.88 7.84 0.06 13.92
(1,118) (1,002) (512) (97) (4,560) (1,366) (2,253) (17) (3,836)
RR (95% CI) 0.92 0.93 0.84 0.80 0.95 1.12 1.33 1.06 1.09
(0.85–0.99) (0.86–1.02) (0.74–0.94) (0.61–1.05) (0.90–0.99) (1.03–1.22) (1.24–1.43) (0.53–2.14) (1.03–1.15)
p Value 0.0422 0.1195 0.0028 0.1064 0.0224 0.0107 <0.0001 0.8655 0.0011
Values are percent (n). CI ¼ confidence interval;
RR ¼ relative risk.
In a recent randomized trial [6], the long-term survival impact of SA was negligible, but observational studies with larger sample sizes and longer follow-up have demonstrated improved survival after SA in normal sinus rhythm [16, 18, 22, 23]. In a study of SA versus noSA, risk-adjusted survival differences were assessed in 372 propensity matched pairs [23]. At 47 months of follow-up, 78% of SA patients were free of AF, and restoration of normal sinus rhythm improved survival. Similar institutional results were published for patients with paroxysmal AF [24]. In a recent administrative study of US Medicare patients undergoing CABG, SA for AF reduced the long-term hazard of death by 42% and resulted in fewer hospital readmissions [25]. Evidence supporting improved survival after SA is becoming more consistent [12, 15, 16, 22, 24]. Nevertheless, surgeons remain concerned about potential added risk of SA, especially for sicker patients. To date, large-scale data on mortality risk of SA remain sparse. In a previous STS ACSD study [11], SA did not add mortality to mitral operations. Those data were acquired 10 years ago, and more recent information has suggested mortality reduction with SA [12]. In the present study, unadjusted outcomes were worse in AF patients not having SA, likely from worse baseline risk. However, after risk adjustment, an independent beneficial effect of SA on mortality was clearly detected. These operative mortality benefits were likely mediated through fewer AF-mediated complications, and results may have improved because of refinement and consistency in the application of SA technology since the last STS evaluation [11]. The current large propensitymatched sample revealed that contemporary SA was indeed safe, and concomitant SA reduced relative risk for mortality by 8% (p ¼ 0.0422), and stroke by 16% (p ¼ 0.0028). Although risk-adjusted mortality, stroke, and prolonged ventilation were reduced in this study, renal failure and new pacemakers were increased after SA (Table 4). The relative risk of renal failure was 1.12, which one may surmise could result from longer operative times associated with SA. It is difficult to determine whether increase in
pacemaker rates represent a surrogate for complications of SA (ie, heart block), as pacemakers may be used as part of primary AF therapy. Distinct from heart block, pacemakers after SA are used for amiodarone-induced bradycardia, overdrive atrial pacing, and treatment of underlying sick sinus syndrome revealed only after correction of persistent AF. The pacemaker issue deserves further study, especially to define clinical reasons for insertion. The surgeon contemplating adding SA to treat AF at the time of cardiac operation will therefore need to reconcile these two aspects: reduced risk of mortality and stroke versus incremental pacemaker risk. Future regression analyses focused on procedural categories and specific predictors may be useful to tease out the nuances of outcomes at the peripheries of patient risk. The recent STS guidelines outline that concomitant SA to restore normal sinus rhythm during mitral valve procedures is a Class 1 recommendation, level of evidence A [26]. Similarly, SA at the time of isolated AVR and AVR with CABG is a Class 1 recommendation, level of evidence B-NR. As the surgical community potentially embraces renewed interest in the impact of SA on short-term outcomes, it must be emphasized that updated long-term outcomes will be important to enhance our understanding. Either through future randomized trials or by incorporating registry data linked to longitudinal data such as available through Medicare, contemporary long-term outcomes of SA will further refine future decision making.
Study Limitations Limitations include possible unmeasured confounders or nonrepresentative populations, as in any observational registry study. The STS ACSD includes all operations from more than 90% of the centers in the United States, and therefore is highly representative of overall US practice. An important limitation in this study was the amount of missing AF type documented. Although the use of inverse probability weighting or other methodologies may be able to statistically adjust for the missing AF documentation, we believe that the existing and wellvalidated STS risk model methodology was most appropriate for this study’s objective. That is especially the case
Ann Thorac Surg 2017;104:493–500
This research was funded by The Society of Thoracic Surgeons’ Access & Publications research program.
References 1. January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: executive summary: a report of the American College of Cardiology/American Heart Association task force on practice guidelines and the Heart Rhythm Society. Circulation 2014;130:2071–104. 2. Calkins H, Kuck KH, Cappato R, et al. 2012 HRS/EHRA/ECAS expert consensus statement on catheter and surgical ablation of atrial fibrillation: recommendations for patient selection, procedural techniques, patient management and follow-up, definitions, endpoints, and research trial design: a report of the Heart Rhythm Society (HRS) task force on catheter and surgical ablation of atrial fibrillation. Heart Rhythm 2012;9:632–96.e21. 3. Saxena A, Dinh DT, Reid CM, Smith JA, Shardey GC, Newcomb AE. Does preoperative atrial fibrillation portend a poorer prognosis in patients undergoing isolated aortic valve replacement? A multicentre Australian study. Can J Cardiol 2013;29:697–703.
499
4. Quader MA, McCarthy PM, Gillinov AM, et al. Does preoperative atrial fibrillation reduce survival after coronary artery bypass grafting? Ann Thorac Surg 2004;77:1514–24. 5. Levy F, Rusinaru D, Marechaux S, Charles V, Peltier M, Tribouilloy C. Determinants and prognosis of atrial fibrillation in patients with aortic stenosis. Am J Cardiol 2015;116:1541–6. 6. Gillinov AM, Gelijns AC, Parides MK, et al. Surgical ablation of atrial fibrillation during mitral-valve surgery. N Engl J Med 2015;372:1399–409. 7. Chevalier P, Leizorovicz A, Maureira P, et al. Left atrial radiofrequency ablation during mitral valve surgery: a prospective randomized multicentre study (SAFIR). Arch Cardiovasc Dis 2009;102:769–75. 8. Blomstrom-Lundqvist C, Johansson B, Berglin E, et al. A randomized double-blind study of epicardial left atrial cryoablation for permanent atrial fibrillation in patients undergoing mitral valve surgery: the Swedish Multicentre Atrial Fibrillation study (SWEDMAF). Eur Heart J 2007;28: 2902–8. 9. Abreu Filho CA, Lisboa LA, Dallan LA, et al. Effectiveness of the maze procedure using cooled-tip radiofrequency ablation in patients with permanent atrial fibrillation and rheumatic mitral valve disease. Circulation 2005;112 (Suppl):I20–5. 10. Doukas G, Samani NJ, Alexiou C, et al. Left atrial radiofrequency ablation during mitral valve surgery for continuous atrial fibrillation: a randomized controlled trial. JAMA 2005;294:2323–9. 11. Gammie JS, Haddad M, Milford-Beland S, et al. Atrial fibrillation correction surgery: lessons from The Society of Thoracic Surgeons National Cardiac Database. Ann Thorac Surg 2008;85:909–14. 12. Rankin JS, He X, O’Brien SM, et al. The Society of Thoracic Surgeons risk model for operative mortality after multiple valve surgery. Ann Thorac Surg 2013;95:1484–90. 13. Lall SC, Melby SJ, Voeller RK, et al. The effect of ablation technology on surgical outcomes after the Cox-maze procedure: a propensity analysis. J Thorac Cardiovasc Surg 2007;133:389–96. 14. Gillinov AM, Bhavani S, Blackstone EH, et al. Surgery for permanent atrial fibrillation: impact of patient factors and lesion set. Ann Thorac Surg 2006;82:502–13. 15. Phan K, Xie A, La Meir M, Black D, Yan TD. Surgical ablation for treatment of atrial fibrillation in cardiac surgery: a cumulative meta-analysis of randomised controlled trials. Heart 2014;100:722–30. 16. Ad N, Holmes SD, Pritchard G, Shuman DJ. Association of operative risk with the outcome of concomitant Cox maze procedure: a comparison of results across risk groups. J Thorac Cardiovasc Surg 2014;148:3027–33. 17. Ad N, Suri RM, Gammie JS, Sheng S, O’Brien SM, Henry L. Surgical ablation of atrial fibrillation trends and outcomes in North America. J Thorac Cardiovasc Surg 2012;144:1051–60. 18. Damiano RJ, Schwartz FH, Bailey MS, et al. The Cox maze IV procedure: predictors of late recurrence. J Thorac Cardiovasc Surg 2011;141:113–21. 19. Ad N, Holmes SD, Shuman DJ, Pritchard G. Impact of atrial fibrillation duration on the success of first-time concomitant Cox maze procedures. Ann Thorac Surg 2015;100:1613–8. 20. Ad N, Henry L, Hunt S, Holmes SD. Do we increase the operative risk by adding the Cox maze III procedure to aortic valve replacement and coronary artery bypass surgery? J Thorac Cardiovasc Surg 2012;143:936–44. 21. Ad N, Holmes SD, Lamont D, Shuman DJ. Left-sided surgical ablation for patients with atrial fibrillation who are undergoing concomitant cardiac surgical procedures. Ann Thorac Surg 2017;103:58–65. 22. Melo J, Santiago T, Aguiar C, et al. Surgery for atrial fibrillation in patients with mitral valve disease: results at five years from the International Registry of Atrial Fibrillation Surgery. J Thorac Cardiovasc Surg 2008;135:863–9. 23. Lee R, McCarthy PM, Wang EC, et al. Midterm survival in patients treated for atrial fibrillation: a propensity-matched
ADULT CARDIAC
as the sample size of actual AF documentation to outcome is the largest in the literature on this topic to date. A similar limitation pertains to incompletely documented data on atrial location and epicardial versus endocardial lesions. These observations highlight the importance for STS participants to record the particulars of AF type and lesion method when performing SA. To enhance future compliance, STS has emphasized the importance of documenting these variables in the revised version of the ACSD available in 2017. To enable the most accurate assessment between SA and no-SA, only the cohort that had AF type documented was utilized for comparative effectiveness analysis. Clearly, it would have been preferable to include all patients, but given the very large sample of propensity matched pairs (n ¼ 28,739) inclusive of AF type, these results most likely reflect an accurate comparative effectiveness sample for SA. Although hospital or participant-site volumes were accounted for in the matching algorithms, precise surgeon experience could not be assessed. A known STS ACSD limitation is lack of echocardiographic data, and left atrial appendage obliteration efficacy was unknown. Although observational results are always subject to confounding variables, the inference of the current analysis is that broader SA application may be safe, but data from large clinical registries do not necessarily replace those from randomized clinical investigation. This issue may be further informed by future longitudinal studies. In conclusion, the application of SA to treat patients with AF undergoing cardiac operations in the United States is increasing. Risk-adjusted operative mortality and stroke are reduced by adding SA, whereas renal failure and pacemaker rates are increased. These findings suggest that further growth in SA during cardiac operations may be appropriate, with potential expansion across all risk profiles. The availability of long-term data for contemporary SA will aid in balancing wider application with clinical appropriateness.
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
ADULT CARDIAC
500
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
comparison to patients without a history of atrial fibrillation. J Thorac Cardiovasc Surg 2012;143:1341–51. 24. McCarthy PM, Manjunath A, Kruse J, et al. Should paroxysmal atrial fibrillation be treated during cardiac surgery? J Thorac Cardiovasc Surg 2013;146:810–23. 25. Rankin JS, Lerner DJ, Braid-Forbes MJ, et al. One-year mortality and costs after surgical ablation for atrial
Ann Thorac Surg 2017;104:493–500
fibrillation concomitant to coronary artery bypass grafting. Eur J Cardiothorac Surg 2017 May 2; [E-Pub ahead of print]. 26. Badhwar V, Rankin JS, Damino RJ, et al. The Society of Thoracic Surgeons 2016 clinical practice guidelines for the surgical treatment of atrial fibrillation. Ann Thorac Surg 2017;103:329–41.
DISCUSSION DR RAVI GHANTA (Charlottesville, VA): I am substituting for Dr Aliwadi as he could not attend today. Vinay, I congratulate you and your coauthors on an insightful study of using The Society of Thoracic Surgeons (STS) database to evaluate contemporary practice patterns of surgical ablation in patients who present for cardiac surgery. Your study shows that surgeons are increasingly performing ablations. Using a propensity matched comparison of 57,478 patients, you importantly found that ablation patients had a decreased 30-day operative mortality and risk of stroke, albeit with an increased risk of renal failure, need for pacemaker, and readmission. As you know, earlier today a similar study was presented using the Virginia STS data, which also found similar outcomes. The Virginia study noted decreased operative mortality and reduced risk of stroke with an increased risk of pacemaker. I have three questions. First, I find it very interesting that in both studies presented today, unadjusted and propensity matched operative mortality and stroke are reduced despite undergoing presumably longer, more complex operations. Do you really believe this is true or are there factors unaccounted for in the propensity matching that may have biased these results? If it is true, can you surmise why operative mortality and stroke may have been reduced with ablation? DR BADHWAR: That is a good question. I believe this to be real. Examining 28,000 comprehensively matched pairs in the creation of the propensity model, we accounted for all available variables, including age and renal failure. We did exclude redo operations from this analysis because it would be an unfair comparison. Overall, this accounted for all the STS major comorbid risk factors. And so I believe that by adding ablation to these patients that are often sicker, the time investment pays outcome dividends. DR GHANTA: Also in this study, roughly half of the patients had paroxsymal atrial fibrillation (AF) and the other half had persistent AF. Did you perform any subgroup analyses that distinguished these two groups? Was there any difference in ablation patterns and left atrial appendage closure dependent on the type of preoperative AF? DR BADHWAR: So there are two questions in that one. They are both very good. The AF type, paroxysmal and persistent, as recorded in the database did go into the propensity model. Patients were indeed matched by the AF type. Therefore, the results are reflective of preoperative AF status. In terms of the left atrial appendage obliteration technique, that was quasidocumented. The obliteration method in the STS database, as you may recall, is just obliteration or not, not necessarily the detail of the technique. The future STS database version will likely describe this. DR GHANTA: And finally, in this study 48% of patients overall with AF underwent concomitant ablation, indicating
underutilization. As you are an author of the STS guidelines on surgical ablation for atrial fibrillation, based on these data and the data presented earlier today and the recent CTSNet trial published in the New England Journal of Medicine, which patients do you recommend undergo concomitant ablation and what lesion set should be performed? What should be the goal and how do we increase utilization? DR BADHWAR: That is an insightful comment and question. It is true that you will soon see the STS guidelines for surgical ablation appear in The Annals. This was written by a panel of multiple experts evaluating the literature, and it will reflect the expanding safety and efficacy of surgical ablation in the literature. For example, there will be a class IA indication for performance of surgical ablation for concomitant mitral surgery as well as other concomitant procedures. To answer your question on patient selection, I think the data are now clear across the board that when patients present with AF for a cardiac operation, surgical ablation may confer an outcome benefit. That being said, guidelines are guidelines, and should be considered as just a suggestion for the surgeon to use their clinical judgment of safety based on one’s own experience. But that being said, barring some significant clinical reason to the contrary, I think the data are now building that any time a patient presents in AF, a surgical ablation should be performed. As to the lesion set, that was not specifically studied in this data set, and really the specificity of lesion set documentation is also lacking. This is a plea today, and in the future, that all participants and surgeons should please record what they are doing when they do it, across all procedures, particularly in AF ablation, because it helps us answer your question. However, the literature does indicate that when surgical ablation is being performed, the most evidence and the most longitudinal outcomes studied is with the full Cox-maze IV lesion set. DR JOSEPH ARCIDI (Flint, MI): You said that the duration and type of AF was accounted for. Was there equivalence for paroxysmal and persistent types, or were there different results with regard to mortality and morbidity for both persistent and paroxysmal varieties, compared with not addressing the AF? DR BADHWAR: So if I would rephrase your question, I think you are asking, was there an outcome differential between paroxysmal and persistent patients. In the comparative effectiveness or the propensity matched model, we included the different subtypes of AF in determining outcome. We did not specifically break them apart to then subset propensity match them, so I do not have that answer for you. However, we do know that in the database when one records paroxysmal, it is the AF documented at the time of surgery in the 2.73 model that we used. So it may or may not reflect true paroxysmal, because one could be on medication, and so forth. So a paroxysmal could be actually a persistent. In the future iterations of the data set, we will hopefully clarify that.
Vinay Badhwar, MD, J. Scott Rankin, MD, Niv Ad, MD, Maria Grau-Sepulveda, MD, MPH, Ralph J. Damiano, MD, A. Marc Gillinov, MD, Patrick M. McCarthy, MD, Vinod H. Thourani, MD, Rakesh M. Suri, MD, DPhil, Jeffrey P. Jacobs, MD, and James L. Cox, MD Department of Cardiovascular and Thoracic Surgery, West Virginia University, Morgantown, West Virginia; Duke Clinical Research Institute, Durham, North Carolina; Division of Cardiothoracic Surgery, Washington University, St. Louis, Missouri; Department of Thoracic and Cardiovascular Surgery, Cleveland Clinic, Cleveland, Ohio; Division of Cardiac Surgery, Northwestern University Feinberg School of Medicine, Chicago, Illinois; Division of Cardiothoracic Surgery, Emory University, Atlanta, Georgia; and Division of Cardiac Surgery, Johns Hopkins University School of Medicine, Baltimore, Maryland
Background. Surgical ablation (SA) for atrial fibrillation (AF) effectively restores sinus rhythm. Incompletely defined risk has previously limited concomitant performance of SA during cardiac operations. The study goals were to define performance trends and risk-adjusted outcomes for contemporary SA. Methods. From July 2011 to June 2014, 86,941 patients with AF, but without endocarditis, underwent primary nonemergent cardiac operations in The Society of Thoracic Surgeons (STS) database. Cochran-Armitage tests examined performance trends of SA for six operative categories: mitral valve repair or replacement (MVRR) with or without coronary artery bypass graft surgery (CABG), aortic valve replacement (AVR) with or without CABG, CABG, AVR with MVRR, stand-alone SA, and other concomitant operations. The risk of concomitant SA was analyzed by propensity matching 28,739 patient-pairs with and without SA by AF type, primary operation, and STS comorbid risk variables using greedy 1:1 matching algorithms. Results. Among all patients with AF, 48.3% (42,066 of 86,941) underwent SA. Mitral operations had the highest
trial fibrillation (AF) at the time of cardiac operations negatively affects 30-day outcome and survival [1–5]. Surgical ablation (SA) restores normal sinus rhythm and improves long-term quality of life [6–10]. Prior registry trends revealed that adoption of SA concomitant to cardiac operations was limited, and concerns over added
A
rate of SA (MVRR ± CABG 68.4% [14,693 of 21,496]; MVRR D AVR 59.1% [1,626 of 2,750]). The AVR ± CABG and isolated CABG rates were 39.3% (6,816 of 17,349) and 32.8% (9,156 of 27,924), respectively. Nearly half of other concomitant operations underwent SA, 47.6% (6,939 of 14,586). Performance frequency increased throughout the study period. After propensity matching, SA was associated with a reduction in relative risk (RR) of 30-day mortality (RR 0.92, 95% confidence interval [CI]: 0.85 to 0.99) and stroke (RR 0.84, 95% CI: 0.74 to 0.94), but an increase in renal failure (RR 1.12, 95% CI: 1.03 to 1.22) and pacemaker implantation (RR 1.33, 95% CI: 1.24 to 1.43). Conclusions. Contemporary utilization of SA is increasing across all operative categories. Performance of SA is accompanied by a 30-day reduction in mortality and stroke. These findings further refine our understanding of the role of SA in the treatment of AF. (Ann Thorac Surg 2017;104:493–500) Ó 2017 by The Society of Thoracic Surgeons
Dr Rankin discloses a financial relationship with AtriCure and Medtronic; Dr Ad with Medtronic, Atricure, LivaNova, Nido Surgical, and Left Atrial Appendage Occlusion LLC; Dr Gillinov with AtriCure, Medtronic, Abbott, St. Jude, Edwards, and Clearflow; Drs Damiano and Cox with AtriCure.
Accepted for publication May 5, 2017. Presented at the Sixty-third Annual Meeting of the Southern Thoracic Surgical Association, Naples, FL, Nov 9–12, 2016. Address correspondence to Dr Badhwar, Department of Cardiovascular and Thoracic Surgery, West Virginia University, 1 Medical Center Dr, Morgantown, WV 26506-8059; email: [email protected].
Ó 2017 by The Society of Thoracic Surgeons Published by Elsevier Inc.
The Supplemental Table and Figure can be viewed in the online version of this article [http://dx.doi. org/10.1016/j.athoracsur.2017.05.016] on http://www. annalsthoracicsurgery.org.
0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2017.05.016
ADULT CARDIAC
Surgical Ablation of Atrial Fibrillation in the United States: Trends and Propensity Matched Outcomes
494
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
ADULT CARDIAC
Abbreviations and Acronyms ACSD AF AVR CABG CI LA MVRR RF RR SA STS
= = = = = = = = = = =
Adult Cardiac Surgery Database atrial fibrillation aortic valve replacement coronary artery bypass graft surgery confidence interval left atrium mitral valve repair or replacement radiofrequency relative risk surgical ablation The Society of Thoracic Surgeons
operative risk may have influenced clinical decision making [11]. Although techniques for SA and experience have improved in recent years, contemporary trends of SA remain incompletely defined. Moreover, for AF patients, recent information suggests operative mortality after cardiac surgery may be improved with SA [12]. The goal of this study was to define performance trends and comparatively assess outcomes of contemporary SA using The Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database (ACSD).
Material and Methods Patient Populations From July 1, 2011, to June 30, 2014, 837,978 cardiac operations were recorded in the STS ACSD; of those, 112,401 (13.4%) had preoperative AF documented. Exclusions were emergent status, endocarditis, reoperation, mechanical circulatory support, transplant, aortic dissection, and pulmonary thromboembolectomy. After exclusions, three populations were studied to examine three objectives: (1) all patients receiving SA were examined for technical procedural trends of SA over time, regardless of AF type documentation; (2) all patients with documented AF type were examined to descriptively assess patients treated with SA or not treated with SA (no-SA); and (3) the risk of adding concomitant SA to a primary cardiac operation was assessed by comprehensive propensity matched comparisons inclusive of AF type, primary operation, and comorbidities, after excluding stand-alone operations owing to lack of a no-SA comparator (Fig 1). In a first trends analysis, all 62,025 SA patients were utilized to document overall performance trends and energy source utilization over time in six operative categories: isolated coronary artery bypass graft surgery (CABG), mitral valve repair or replacement (MVRR) with or without CABG, aortic valve replacement (AVR) with or without CABG, AVR with or without MVRR, stand-alone SA, and other concomitant operations. A second comparative descriptive analysis of SA versus no-SA included all 86,941 patients with documented AF type, after operative exclusions. These were assessed in total and by each of the six operative categories. Finally, a third comparative outcomes analysis assessed the risk of
Ann Thorac Surg 2017;104:493–500
adding SA as a concomitant procedure. This analysis included all the above concomitant operations by excluding the 2,836 stand-alone SA patients. The resulting cohort of 84,105 patients with documented AF was evaluated by a propensity matched analysis of concomitant SA versus no-SA.
Baseline Characteristics All relevant procedural STS covariates and baseline characteristics were evaluated using medians with 25th and 75th percentiles for continuous variables and proportions for categoric variables. Descriptive data were compared between groups using the Wilcoxon rank sum test for continuous and ordinal variables, and a Pearson c2 test or Fisher’s exact test for categoric variables, as appropriate.
Trends Analysis Temporal trends in SA utilization across the six operative categories were evaluated with Cochran-Armitage trends tests. The 3 years of data were divided into six 6-month sextiles: period 1 (July 1, 2011, to December 31, 2011) to period 6 (January 1, 2014, to June 30, 2014). Total numbers of procedures and incidence of SA for each procedural type were assessed.
Propensity-Matched Comparative Outcome Analysis To quantify risk-adjusted outcome differences for SA versus no-SA with documented preoperative AF type, a comprehensive propensity matched analysis was performed. Using multivariate logistic regression, probability of treatment assignment to SA versus no-SA (propensity score) was generated from observed covariates and procedures. Missing values (less than 3%) underwent simple imputation in accordance with validated STS models [12]. Group-specific medians were examined for continuous variables, and most common category for categoric variables. Continuous variables were evaluated using restricted cubic spline plots to assess linearity. The final propensity model included the following: AF type (paroxysmal versus persistent), age (linear splines with knot at 75 years), sex, body surface area, creatinine (linear splines with knot at 1.00), ejection fraction (linear splines with knot at 50%), body surface area (quadratic spline), multiplicative interaction between body surface area and sex, New York Heart Association class III to IV, cerebrovascular disease, stroke, chronic lung disease, peripheral vascular disease, home oxygen, sleep apnea, syncope, diabetes type, preoperative myocardial infarction, operative status, previous cardiovascular interventions, previous pacemaker, mitral and tricuspid insufficiency, and operative procedure type (CABG, MVRR CABG, AVR CABG, AVR þ MVRR, other concomitant operations). Propensity scores matched SA and no-SA using a 1:1 greedy 5-to-1 digitmatching algorithm. Only pairs matched on three or more propensity score digits were retained in the final matched sample of SA (n ¼ 28,739) and no-SA (n ¼ 28,739; Supplemental Table). Propensity score distribution and protocol adequacy were assessed to a uniform
Ann Thorac Surg 2017;104:493–500
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
495
1% standardized difference or less between matched pairs (Supplemental Figure). Outcome risk ratios and 95% confidence intervals (CI) for SA versus no-SA were generated for unadjusted and propensity matched comparisons, using generalized estimating equation–modified Poisson regression models with a log-link and an independent correlation structure to account for hospital or participant site clustering. All analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC). A p value less than 0.05 was considered significant. This study of deidentified data from the STS ACSD has been granted a waiver of informed consent by the Duke University Institutional Review Board.
Results Procedural Trends of Surgical Ablation Overall volume of SA showed incremental growth progressively through each of the 6-month sextiles as well as an overall increase of 50% during the study period (p < 0.0001). When analyzed by operation category, significant upward trends in SA rates were observed for all groups except the two least frequent operations (AVR þ MVRR and stand alone; Fig 2). Procedural details are outlined in Table 1. Among all 62,025 SA operations, the ablation location was recorded as left atrium (LA) only in 52.9%, biatrial in 35.8%, 1.7% right atrium only, and 9.6% missing. The left atrial appendage was surgically obliterated in 86.4% overall. Radiofrequency (RF) ablation was used in 49.2% of all
cases, cryoablation in 27.3%, and cut-and-sew lesions applied in 13.1%. Only 2.5% of all cases had ultrasound, microwave, or laser energy applied. Energy source type was missing in 7.9%. Examining specific ablation strategies applied, RF only was used in 36.5%, cryoablation only in 15.3%, cut-and-sew lesions only in 8.4%, and RF plus cryoablation combination in 9.2%, with other combinations being rare.
Contemporary Performance of Surgical Ablation for Atrial Fibrillation In patients with documented AF (n ¼ 86,941), the overall incidence of SA was 48.3%. In this cohort, the most common operation performed was isolated CABG (33.1%), but that had the lowest rate of concomitant SA (33.0%). A MVRR with or without CABG was performed in 25.0%, but had the highest rate of SA (68.4%). The frequency of concomitant SA in AVR with MVRR was 59.1%, other concomitant procedures was 47.6%, and AVR with or without CABG was 39.3%. Baseline characteristics were generally better in SA patients, but the incidences of severe mitral regurgitation and permanent AF were higher (Table 2).
Comparative Outcome Analysis With or Without Concomitant Surgical Ablation Unadjusted operative mortality and most postoperative morbidities were lower with SA, except for permanent pacemaker rates, which were higher in SA versus no-SA (7.6% versus 4.4%, respectively, p < 0.0001). Readmission rates within 30 days were similar (Table 3).
ADULT CARDIAC
Fig 1. Flow diagram of study populations using Consolidated Standards of Reporting Trials (CONSORT) principles. Source: The Society of Thoracic Surgeons Adult Cardiac Database, July 1, 2011, through June 30, 2014. (AF ¼ atrial fibrillation.)
ADULT CARDIAC
496
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
Ann Thorac Surg 2017;104:493–500
Fig 2. Trends in total numbers of surgical ablation procedures in the United States, July 1, 2011, through June 30, 2014. (AVR ¼ aortic valve replacement; CABG ¼ coronary artery bypass graft surgery; MVRR ¼ mitral valve repair or replacement.)
Analyzing 28,739 propensity matched pairs revealed that SA had lower risk-adjusted mortality (RR 0.92, 95% CI: 0.85 to 0.99), permanent stroke (RR 0.84, 95% CI: 0.74 to 0.94), and prolonged ventilation (RR 0.95, 95% CI: 0.90 to 0.99). Pacemaker implantation was more prevalent in the SA group (RR 1.33, 95% CI: 1.24 to 1.43), as was postoperative renal failure (RR 1.12, 95% CI: 1.03 to 1.22) and overall 30-day readmissions (RR 1.09, 95% CI: 1.03 to 1.15). Phrenic nerve injury remained very rare (0.06%) and was equal between groups (Table 4).
Comment This study illustrates the rising utilization of SA concomitant to cardiac operations in the United States. Importantly, by examining 57,478 propensity-matched
patients, the performance of SA was protective of stroke, prolonged ventilation, and 30-day operative mortality, but was associated with an increase in pacemaker risk and perioperative renal failure. The benefit of SA is to restore normal sinus rhythm and quality of life, especially when performed concomitant to mitral surgery [6, 9–11, 13–16]. Nevertheless, registry investigations at earlier sampling periods did not document widespread adoption [11, 17]. A 2005 to 2010 STS ACSD series noted slight decreasing trends over time, with only 40.6% of AF patients undergoing SA [17]. Within the past 5 years, clinical interest in standardized SA techniques have grown substantially [18–20]. These correlate with the Food and Drug Administration– approved indication of SA for AF with one device manufacturer. At mitral operation, the exposed LA
Table 1. Surgical Ablation Procedures Performed by Operation Type
Variable Left atrium only location Biatrial location Atrial location not documented Primarily endocardial Primarily epicardial Ablation location not documented Radiofrequency only Cryoablation only Cut-and-sew lesions only Radiofrequency and cryoablation Left atrial appendage obliterated
Isolated Other Stand MVRR CABG AVR CABG AVR þ MVRR CABG Concomitant Alone (n ¼ 21,992) (n ¼ 9,875) (n ¼ 2,304) (n ¼ 14,334) (n ¼ 10,252) (n ¼ 3,268) 50.9 40.0 9.1 52.4 31.2 16.4
57.9 30.7 11.4 26.6 56.3 17.1
52.1 37.9 10.0 49.3 35.6 15.1
58.0 29.3 12.7 22.0 55.8 22.2
50.5 38.1 11.4 39.9 43.4 16.7
37.2 56.3 6.5 23.6 72.5 3.9
<0.0001
27.0 23.8 7.6 11.8 87.7
43.1 8.9 8.6 7.0 88.5
28.8 19.5 9.0 11.8 86.3
42.5 7.4 10.2 5.5 88.6
34.0 15.1 9.5 9.7 85.9
67.8 9.2 0.8 11.0 63.9
<0.0001
Values are percentages. Surgical ablation techniques performed between July 1, 2011, and June 30, 2014. AVR ¼ atrial valve replacement;
p Value
CABG ¼ coronary artery bypass graft surgery;
MVRR ¼ mitral valve repair or replacement.
<0.0001
<0.0001
Ann Thorac Surg 2017;104:493–500
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
497
Table 2. Patient Characteristics in the Overall Atrial Fibrillation Population Variable
No Ablation (n ¼ 44,875)
Ablation (n ¼ 42,066)
p Value
72 (65–79) 35.0 55 (43–60) 18.2 2.8 15.0 13.4 9.4 16.6 32.0 47.9 50.7 25.2
73 (65–80) 33.6 55 (40–60) 19.6 3.9 17.1 16.1 11.0 23.0 33.4 41.6 58.4 16.9
71 (64–78) 36.7 55 (45–60) 16.7 1.7 12.7 10.6 7.7 9.8 30.5 54.6 45.4 33.9
<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001
Age, years Female Ejection fraction Cerebrovascular disease Preoperative dialysis Moderate-severe COPD Peripheral vascular disease Preoperative pacemaker Preoperative MI within 21 days NYHA class III or IV Persistent atrial fibrillation Paroxysmal atrial fibrillation Severe mitral regurgitation
Values are median (interquartile range) or percentage. Unadjusted baseline patient characteristics in patients with atrial fibrillation between July 1, 2011 and June 30, 2014. COPD ¼ chronic obstructive pulmonary disease;
MI ¼ myocardial infarction;
provides an opportune time to perform SA. In the current study, SA was performed in 48.3% of all patients with AF, and in 68% of concomitant MVRR operations—the highest among all operations. Logically, MVRR facilitated the highest frequency of primarily endocardial lesions (Table 1). Conversely, at the time of closed atrial operations such as isolated CABG or AVR with or without CABG, epicardial, RF only, and LA only lesions were most commonly applied. Many would advocate left atrial appendage obliteration for its impact on long-term morbidity reduction and need for prolonged anticoagulation. Although left atrial appendage obliteration was performed in more than 85% of concomitant SA procedures, it was only completed in 63.9% of stand-alone cases, possibly
NYHA ¼ New York Heart Association.
reflective of the recent resurgence of alternative hybrid procedures. This finding may represent an opportunity for improvement. The longitudinal merits of the full biatrial Cox maze IV lesion set, particularly for persistent AF, are known [14, 18, 19]. However, controversy persists, with recent reasonable results of LA-only lesions [6, 21]. Current practice appears to be reflective of this controversy. The most common lesions applied in all concomitant categories was LA only (Table 2). The exception was with stand-alone SA, in which primarily endocardial biatrial lesions prevail. Although these data do not document long-term outcomes, they do highlight an opportunity for further clarity by identifying which patient subsets would benefit most from LA versus biatrial lesions.
Table 3. Unadjusted Outcomes of Atrial Fibrillation Patients With or Without Concomitant Surgical Ablation Variable Operative mortality Reoperation for bleeding Permanent stroke Transient ischemic attack Prolonged ventilation >48 hours Sternal infection Permanent pacemaker Phrenic nerve injury Discharged home Discharged extended care Discharge coumadin Discharge thrombin inhibitor Discharge antiarrhythmic drug Less than 30-day readmission Pacemaker readmission
Overall (n ¼ 84,105)
No Ablation (n ¼ 44,875)
Ablation (n ¼ 39,230)
p Value
4.4 3.6 2.0 0.4 16.5 0.3 5.9 0.5 64.4 28.9 57.8 5.2 53.2 13.4 2.5
4.9 3.6 2.2 0.4 17.7 0.3 4.4 0.4 61.1 31.5 52.7 4.8 50.2 13.4 2.2
3.8 3.6 1.7 0.3 15.1 0.2 7.6 0.6 68.1 25.9 63.6 5.7 56.5 13.4 2.8
<0.0001 0.8736 <0.0001 0.0142 <0.0001 0.1398 <0.0001 0.2287 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.4497 <0.0001
Values are percentages. Concomitant surgical ablation with known atrial fibrillation performed between July 1, 2011, and June 30, 2014, after excluding 2,836 patients with stand-alone therapy.
ADULT CARDIAC
Overall (n ¼ 86,941)
498
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
Ann Thorac Surg 2017;104:493–500
Table 4. Relative Risks of Performing Concomitant Surgical Ablation in Propensity Matched Patients With Atrial Fibrillation During Adult Cardiac Operations Overall (n ¼ 57,478)
ADULT CARDIAC
Outcome Mortality Reoperation for bleeding Permanent stroke Transient ischemic attack Prolonged ventilation >48 hours Renal failure Pacemaker Phrenic nerve injury Readmission 30 days
4.31 3.61 1.96 0.38 16.31 4.62 6.87 0.06 13.36
(2,480) (2,075) (1,124) (218) (9,373) (2,585) (3,946) (33) (7,347)
No Ablation (n ¼ 28,739) 4.50 3.73 2.13 0.42 16.75 4.35 5.89 0.06 12.79
(1,292) (1,073) (612) (121) (4,813) (1,219) (1,693) (16) (3,511)
Ablation (n ¼ 28,739) 4.13 3.49 1.78 0.34 15.87 4.88 7.84 0.06 13.92
(1,118) (1,002) (512) (97) (4,560) (1,366) (2,253) (17) (3,836)
RR (95% CI) 0.92 0.93 0.84 0.80 0.95 1.12 1.33 1.06 1.09
(0.85–0.99) (0.86–1.02) (0.74–0.94) (0.61–1.05) (0.90–0.99) (1.03–1.22) (1.24–1.43) (0.53–2.14) (1.03–1.15)
p Value 0.0422 0.1195 0.0028 0.1064 0.0224 0.0107 <0.0001 0.8655 0.0011
Values are percent (n). CI ¼ confidence interval;
RR ¼ relative risk.
In a recent randomized trial [6], the long-term survival impact of SA was negligible, but observational studies with larger sample sizes and longer follow-up have demonstrated improved survival after SA in normal sinus rhythm [16, 18, 22, 23]. In a study of SA versus noSA, risk-adjusted survival differences were assessed in 372 propensity matched pairs [23]. At 47 months of follow-up, 78% of SA patients were free of AF, and restoration of normal sinus rhythm improved survival. Similar institutional results were published for patients with paroxysmal AF [24]. In a recent administrative study of US Medicare patients undergoing CABG, SA for AF reduced the long-term hazard of death by 42% and resulted in fewer hospital readmissions [25]. Evidence supporting improved survival after SA is becoming more consistent [12, 15, 16, 22, 24]. Nevertheless, surgeons remain concerned about potential added risk of SA, especially for sicker patients. To date, large-scale data on mortality risk of SA remain sparse. In a previous STS ACSD study [11], SA did not add mortality to mitral operations. Those data were acquired 10 years ago, and more recent information has suggested mortality reduction with SA [12]. In the present study, unadjusted outcomes were worse in AF patients not having SA, likely from worse baseline risk. However, after risk adjustment, an independent beneficial effect of SA on mortality was clearly detected. These operative mortality benefits were likely mediated through fewer AF-mediated complications, and results may have improved because of refinement and consistency in the application of SA technology since the last STS evaluation [11]. The current large propensitymatched sample revealed that contemporary SA was indeed safe, and concomitant SA reduced relative risk for mortality by 8% (p ¼ 0.0422), and stroke by 16% (p ¼ 0.0028). Although risk-adjusted mortality, stroke, and prolonged ventilation were reduced in this study, renal failure and new pacemakers were increased after SA (Table 4). The relative risk of renal failure was 1.12, which one may surmise could result from longer operative times associated with SA. It is difficult to determine whether increase in
pacemaker rates represent a surrogate for complications of SA (ie, heart block), as pacemakers may be used as part of primary AF therapy. Distinct from heart block, pacemakers after SA are used for amiodarone-induced bradycardia, overdrive atrial pacing, and treatment of underlying sick sinus syndrome revealed only after correction of persistent AF. The pacemaker issue deserves further study, especially to define clinical reasons for insertion. The surgeon contemplating adding SA to treat AF at the time of cardiac operation will therefore need to reconcile these two aspects: reduced risk of mortality and stroke versus incremental pacemaker risk. Future regression analyses focused on procedural categories and specific predictors may be useful to tease out the nuances of outcomes at the peripheries of patient risk. The recent STS guidelines outline that concomitant SA to restore normal sinus rhythm during mitral valve procedures is a Class 1 recommendation, level of evidence A [26]. Similarly, SA at the time of isolated AVR and AVR with CABG is a Class 1 recommendation, level of evidence B-NR. As the surgical community potentially embraces renewed interest in the impact of SA on short-term outcomes, it must be emphasized that updated long-term outcomes will be important to enhance our understanding. Either through future randomized trials or by incorporating registry data linked to longitudinal data such as available through Medicare, contemporary long-term outcomes of SA will further refine future decision making.
Study Limitations Limitations include possible unmeasured confounders or nonrepresentative populations, as in any observational registry study. The STS ACSD includes all operations from more than 90% of the centers in the United States, and therefore is highly representative of overall US practice. An important limitation in this study was the amount of missing AF type documented. Although the use of inverse probability weighting or other methodologies may be able to statistically adjust for the missing AF documentation, we believe that the existing and wellvalidated STS risk model methodology was most appropriate for this study’s objective. That is especially the case
Ann Thorac Surg 2017;104:493–500
This research was funded by The Society of Thoracic Surgeons’ Access & Publications research program.
References 1. January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: executive summary: a report of the American College of Cardiology/American Heart Association task force on practice guidelines and the Heart Rhythm Society. Circulation 2014;130:2071–104. 2. Calkins H, Kuck KH, Cappato R, et al. 2012 HRS/EHRA/ECAS expert consensus statement on catheter and surgical ablation of atrial fibrillation: recommendations for patient selection, procedural techniques, patient management and follow-up, definitions, endpoints, and research trial design: a report of the Heart Rhythm Society (HRS) task force on catheter and surgical ablation of atrial fibrillation. Heart Rhythm 2012;9:632–96.e21. 3. Saxena A, Dinh DT, Reid CM, Smith JA, Shardey GC, Newcomb AE. Does preoperative atrial fibrillation portend a poorer prognosis in patients undergoing isolated aortic valve replacement? A multicentre Australian study. Can J Cardiol 2013;29:697–703.
499
4. Quader MA, McCarthy PM, Gillinov AM, et al. Does preoperative atrial fibrillation reduce survival after coronary artery bypass grafting? Ann Thorac Surg 2004;77:1514–24. 5. Levy F, Rusinaru D, Marechaux S, Charles V, Peltier M, Tribouilloy C. Determinants and prognosis of atrial fibrillation in patients with aortic stenosis. Am J Cardiol 2015;116:1541–6. 6. Gillinov AM, Gelijns AC, Parides MK, et al. Surgical ablation of atrial fibrillation during mitral-valve surgery. N Engl J Med 2015;372:1399–409. 7. Chevalier P, Leizorovicz A, Maureira P, et al. Left atrial radiofrequency ablation during mitral valve surgery: a prospective randomized multicentre study (SAFIR). Arch Cardiovasc Dis 2009;102:769–75. 8. Blomstrom-Lundqvist C, Johansson B, Berglin E, et al. A randomized double-blind study of epicardial left atrial cryoablation for permanent atrial fibrillation in patients undergoing mitral valve surgery: the Swedish Multicentre Atrial Fibrillation study (SWEDMAF). Eur Heart J 2007;28: 2902–8. 9. Abreu Filho CA, Lisboa LA, Dallan LA, et al. Effectiveness of the maze procedure using cooled-tip radiofrequency ablation in patients with permanent atrial fibrillation and rheumatic mitral valve disease. Circulation 2005;112 (Suppl):I20–5. 10. Doukas G, Samani NJ, Alexiou C, et al. Left atrial radiofrequency ablation during mitral valve surgery for continuous atrial fibrillation: a randomized controlled trial. JAMA 2005;294:2323–9. 11. Gammie JS, Haddad M, Milford-Beland S, et al. Atrial fibrillation correction surgery: lessons from The Society of Thoracic Surgeons National Cardiac Database. Ann Thorac Surg 2008;85:909–14. 12. Rankin JS, He X, O’Brien SM, et al. The Society of Thoracic Surgeons risk model for operative mortality after multiple valve surgery. Ann Thorac Surg 2013;95:1484–90. 13. Lall SC, Melby SJ, Voeller RK, et al. The effect of ablation technology on surgical outcomes after the Cox-maze procedure: a propensity analysis. J Thorac Cardiovasc Surg 2007;133:389–96. 14. Gillinov AM, Bhavani S, Blackstone EH, et al. Surgery for permanent atrial fibrillation: impact of patient factors and lesion set. Ann Thorac Surg 2006;82:502–13. 15. Phan K, Xie A, La Meir M, Black D, Yan TD. Surgical ablation for treatment of atrial fibrillation in cardiac surgery: a cumulative meta-analysis of randomised controlled trials. Heart 2014;100:722–30. 16. Ad N, Holmes SD, Pritchard G, Shuman DJ. Association of operative risk with the outcome of concomitant Cox maze procedure: a comparison of results across risk groups. J Thorac Cardiovasc Surg 2014;148:3027–33. 17. Ad N, Suri RM, Gammie JS, Sheng S, O’Brien SM, Henry L. Surgical ablation of atrial fibrillation trends and outcomes in North America. J Thorac Cardiovasc Surg 2012;144:1051–60. 18. Damiano RJ, Schwartz FH, Bailey MS, et al. The Cox maze IV procedure: predictors of late recurrence. J Thorac Cardiovasc Surg 2011;141:113–21. 19. Ad N, Holmes SD, Shuman DJ, Pritchard G. Impact of atrial fibrillation duration on the success of first-time concomitant Cox maze procedures. Ann Thorac Surg 2015;100:1613–8. 20. Ad N, Henry L, Hunt S, Holmes SD. Do we increase the operative risk by adding the Cox maze III procedure to aortic valve replacement and coronary artery bypass surgery? J Thorac Cardiovasc Surg 2012;143:936–44. 21. Ad N, Holmes SD, Lamont D, Shuman DJ. Left-sided surgical ablation for patients with atrial fibrillation who are undergoing concomitant cardiac surgical procedures. Ann Thorac Surg 2017;103:58–65. 22. Melo J, Santiago T, Aguiar C, et al. Surgery for atrial fibrillation in patients with mitral valve disease: results at five years from the International Registry of Atrial Fibrillation Surgery. J Thorac Cardiovasc Surg 2008;135:863–9. 23. Lee R, McCarthy PM, Wang EC, et al. Midterm survival in patients treated for atrial fibrillation: a propensity-matched
ADULT CARDIAC
as the sample size of actual AF documentation to outcome is the largest in the literature on this topic to date. A similar limitation pertains to incompletely documented data on atrial location and epicardial versus endocardial lesions. These observations highlight the importance for STS participants to record the particulars of AF type and lesion method when performing SA. To enhance future compliance, STS has emphasized the importance of documenting these variables in the revised version of the ACSD available in 2017. To enable the most accurate assessment between SA and no-SA, only the cohort that had AF type documented was utilized for comparative effectiveness analysis. Clearly, it would have been preferable to include all patients, but given the very large sample of propensity matched pairs (n ¼ 28,739) inclusive of AF type, these results most likely reflect an accurate comparative effectiveness sample for SA. Although hospital or participant-site volumes were accounted for in the matching algorithms, precise surgeon experience could not be assessed. A known STS ACSD limitation is lack of echocardiographic data, and left atrial appendage obliteration efficacy was unknown. Although observational results are always subject to confounding variables, the inference of the current analysis is that broader SA application may be safe, but data from large clinical registries do not necessarily replace those from randomized clinical investigation. This issue may be further informed by future longitudinal studies. In conclusion, the application of SA to treat patients with AF undergoing cardiac operations in the United States is increasing. Risk-adjusted operative mortality and stroke are reduced by adding SA, whereas renal failure and pacemaker rates are increased. These findings suggest that further growth in SA during cardiac operations may be appropriate, with potential expansion across all risk profiles. The availability of long-term data for contemporary SA will aid in balancing wider application with clinical appropriateness.
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
ADULT CARDIAC
500
BADHWAR ET AL SURGICAL ABLATION IN THE UNITED STATES
comparison to patients without a history of atrial fibrillation. J Thorac Cardiovasc Surg 2012;143:1341–51. 24. McCarthy PM, Manjunath A, Kruse J, et al. Should paroxysmal atrial fibrillation be treated during cardiac surgery? J Thorac Cardiovasc Surg 2013;146:810–23. 25. Rankin JS, Lerner DJ, Braid-Forbes MJ, et al. One-year mortality and costs after surgical ablation for atrial
Ann Thorac Surg 2017;104:493–500
fibrillation concomitant to coronary artery bypass grafting. Eur J Cardiothorac Surg 2017 May 2; [E-Pub ahead of print]. 26. Badhwar V, Rankin JS, Damino RJ, et al. The Society of Thoracic Surgeons 2016 clinical practice guidelines for the surgical treatment of atrial fibrillation. Ann Thorac Surg 2017;103:329–41.
DISCUSSION DR RAVI GHANTA (Charlottesville, VA): I am substituting for Dr Aliwadi as he could not attend today. Vinay, I congratulate you and your coauthors on an insightful study of using The Society of Thoracic Surgeons (STS) database to evaluate contemporary practice patterns of surgical ablation in patients who present for cardiac surgery. Your study shows that surgeons are increasingly performing ablations. Using a propensity matched comparison of 57,478 patients, you importantly found that ablation patients had a decreased 30-day operative mortality and risk of stroke, albeit with an increased risk of renal failure, need for pacemaker, and readmission. As you know, earlier today a similar study was presented using the Virginia STS data, which also found similar outcomes. The Virginia study noted decreased operative mortality and reduced risk of stroke with an increased risk of pacemaker. I have three questions. First, I find it very interesting that in both studies presented today, unadjusted and propensity matched operative mortality and stroke are reduced despite undergoing presumably longer, more complex operations. Do you really believe this is true or are there factors unaccounted for in the propensity matching that may have biased these results? If it is true, can you surmise why operative mortality and stroke may have been reduced with ablation? DR BADHWAR: That is a good question. I believe this to be real. Examining 28,000 comprehensively matched pairs in the creation of the propensity model, we accounted for all available variables, including age and renal failure. We did exclude redo operations from this analysis because it would be an unfair comparison. Overall, this accounted for all the STS major comorbid risk factors. And so I believe that by adding ablation to these patients that are often sicker, the time investment pays outcome dividends. DR GHANTA: Also in this study, roughly half of the patients had paroxsymal atrial fibrillation (AF) and the other half had persistent AF. Did you perform any subgroup analyses that distinguished these two groups? Was there any difference in ablation patterns and left atrial appendage closure dependent on the type of preoperative AF? DR BADHWAR: So there are two questions in that one. They are both very good. The AF type, paroxysmal and persistent, as recorded in the database did go into the propensity model. Patients were indeed matched by the AF type. Therefore, the results are reflective of preoperative AF status. In terms of the left atrial appendage obliteration technique, that was quasidocumented. The obliteration method in the STS database, as you may recall, is just obliteration or not, not necessarily the detail of the technique. The future STS database version will likely describe this. DR GHANTA: And finally, in this study 48% of patients overall with AF underwent concomitant ablation, indicating
underutilization. As you are an author of the STS guidelines on surgical ablation for atrial fibrillation, based on these data and the data presented earlier today and the recent CTSNet trial published in the New England Journal of Medicine, which patients do you recommend undergo concomitant ablation and what lesion set should be performed? What should be the goal and how do we increase utilization? DR BADHWAR: That is an insightful comment and question. It is true that you will soon see the STS guidelines for surgical ablation appear in The Annals. This was written by a panel of multiple experts evaluating the literature, and it will reflect the expanding safety and efficacy of surgical ablation in the literature. For example, there will be a class IA indication for performance of surgical ablation for concomitant mitral surgery as well as other concomitant procedures. To answer your question on patient selection, I think the data are now clear across the board that when patients present with AF for a cardiac operation, surgical ablation may confer an outcome benefit. That being said, guidelines are guidelines, and should be considered as just a suggestion for the surgeon to use their clinical judgment of safety based on one’s own experience. But that being said, barring some significant clinical reason to the contrary, I think the data are now building that any time a patient presents in AF, a surgical ablation should be performed. As to the lesion set, that was not specifically studied in this data set, and really the specificity of lesion set documentation is also lacking. This is a plea today, and in the future, that all participants and surgeons should please record what they are doing when they do it, across all procedures, particularly in AF ablation, because it helps us answer your question. However, the literature does indicate that when surgical ablation is being performed, the most evidence and the most longitudinal outcomes studied is with the full Cox-maze IV lesion set. DR JOSEPH ARCIDI (Flint, MI): You said that the duration and type of AF was accounted for. Was there equivalence for paroxysmal and persistent types, or were there different results with regard to mortality and morbidity for both persistent and paroxysmal varieties, compared with not addressing the AF? DR BADHWAR: So if I would rephrase your question, I think you are asking, was there an outcome differential between paroxysmal and persistent patients. In the comparative effectiveness or the propensity matched model, we included the different subtypes of AF in determining outcome. We did not specifically break them apart to then subset propensity match them, so I do not have that answer for you. However, we do know that in the database when one records paroxysmal, it is the AF documented at the time of surgery in the 2.73 model that we used. So it may or may not reflect true paroxysmal, because one could be on medication, and so forth. So a paroxysmal could be actually a persistent. In the future iterations of the data set, we will hopefully clarify that.