Advertisement

De novo atrial fibrillation after mitral valve surgery

  • Christopher K. Mehta
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • Patrick M. McCarthy
    Correspondence
    Address for reprints: Patrick M. McCarthy, MD, Division of Cardiac Surgery, Northwestern University Feinberg School of Medicine and Northwestern Memorial Hospital, Bluhm Cardiovascular Institute, 201 E Huron St, Suite 11-140, Chicago, IL 60611.
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • Adin-Cristian Andrei
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • Jane Kruse
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • Hangzhi Shi
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • Andrei Churyla
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • S. Chris Malaisrie
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • Duc Thinh Pham
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
  • James L. Cox
    Affiliations
    Division of Cardiac Surgery, Department of Surgery, Northwestern University Feinberg School of Medicine and Northwestern Medicine, Bluhm Cardiovascular Institute, Chicago, Ill
    Search for articles by this author
Open ArchivePublished:April 18, 2018DOI:https://doi.org/10.1016/j.jtcvs.2018.04.050

      Abstract

      Objectives

      We sought to determine the incidence and risk factors for de novo atrial fibrillation (>90 days after surgery) in patients without preoperative atrial fibrillation.

      Methods

      From 2004 to 2014, 2261 patients underwent mitral valve surgery; 1288 patients (57%) did not have a history of atrial fibrillation, and 930 patients had rhythm information more than 90 days after surgery. De novo atrial fibrillation and death probabilities were estimated using a semi-competing risks, multi-state model. Univariable and multivariable risk factors for developing atrial fibrillation were identified using the Fine–Gray model.

      Results

      The 5- and 10-year incidences of de novo atrial fibrillation were 14% and 23%, respectively. Univariable risk factors were older age, more complex operations, more tricuspid regurgitation, and congestive heart failure (all P < .05). Patients with degenerative mitral regurgitation were less likely to develop atrial fibrillation (hazard ratio [HR], 0.4; 95% confidence interval [CI], 0.24-0.65; P < .001). Multivariable risk factors for de novo atrial fibrillation were tricuspid valve surgery (HR, 1.80; 95% CI, 1.22, 2.65; P = .003), aortic valve surgery (HR, 1.49; 95% CI, 1.03-2.17; P = .035), and older age (HR, 1.03; 95% CI, 1.02-1.05; P < .001). De novo atrial fibrillation did not affect overall survival (P = .41). Among patients who developed de novo atrial fibrillation, we observed increased use of warfarin (P < .001) and a strong trend toward an increased risk of stroke (P = .055).

      Conclusions

      De novo atrial fibrillation develops progressively after mitral surgery and is associated with a strong trend toward stroke. Patients at high risk could be studied in a trial to reduce atrial fibrillation.

      Key Words

      Abbreviations and Acronyms:

      AF (atrial fibrillation), AV (aortic valve), CARD (Cardiovascular Research Database), CI (confidence interval), DMR (degenerative mitral regurgitation), HR (hazard ratio), MV (mitral valve), PS (propensity score), TR (tricuspid regurgitation), TV (tricuspid valve)
      Figure thumbnail fx1
      Freedom from stroke by AF status in follow-up.
      A steady rate of de novo AF (>90 days postsurgery) develops after MV surgery in patients without a preoperative history of AF and was associated with a trend (P = .055) toward increased stroke.
      Development of de novo AF after MV surgery may affect the risk of stroke and development of functional TR. Identifying patients at risk of developing AF and treating them with prophylactic ablation may reduce the risk of AF, lower stroke risk, and limit late functional TR.
      See Editorial Commentary page 1526.
      See Editorial page 1514.
      Preoperative atrial fibrillation (AF) in patients undergoing surgery for mitral valve (MV) disease is associated with late adverse events and reduced survival.
      • Lim E.
      • Barlow C.W.
      • Hosseinpour A.R.
      • Wisbey C.
      • Wilson K.
      • Pidgeon W.
      • et al.
      Influence of atrial fibrillation on outcome following mitral valve repair.
      • Ngaage D.L.
      • Schaff H.V.
      • Mullany C.J.
      • Barnes S.
      • Dearani J.A.
      • Daly R.C.
      • et al.
      Influence of preoperative atrial fibrillation on late results of mitral repair: is concomitant ablation justified?.
      When AF is identified preoperatively, surgical ablation is recommended to attempt sinus rhythm restoration.
      • Calkins H.
      • Kuck K.H.
      • Cappato R.
      • Brugada J.
      • Camm A.J.
      • Chen S.A.
      • 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. Developed in partnership with the European Heart Rhythm Association (EHRA), a registered branch of the European Society of Cardiology (ESC) and the European Cardiac Arrhythmia Society (ECAS); and in collaboration with the American College of Cardiology (ACC), American Heart Association (AHA), the Asia Pacific Heart Rhythm Society (APHRS), and the Society of Thoracic Surgeons (STS). Endorsed by the governing bodies of the American College of Cardiology Foundation, the American Heart Association, the European Cardiac Arrhythmia Society, the European Heart Rhythm Association, the Society of Thoracic Surgeons, the Asia Pacific Heart Rhythm Society, and the Heart Rhythm Society.
      • Badhwar V.
      • Rankin J.S.
      • Damiano Jr., R.J.
      • Gillinov A.M.
      • Bakaeen F.G.
      • Edgerton J.R.
      • et al.
      The Society of Thoracic Surgeons 2017 clinical practice guidelines for the surgical treatment of atrial fibrillation.
      Multiple prospective randomized studies have demonstrated that the addition of surgical ablation to MV surgery for patients with preoperative AF significantly increases freedom from AF.
      • Calkins H.
      • Kuck K.H.
      • Cappato R.
      • Brugada J.
      • Camm A.J.
      • Chen S.A.
      • 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. Developed in partnership with the European Heart Rhythm Association (EHRA), a registered branch of the European Society of Cardiology (ESC) and the European Cardiac Arrhythmia Society (ECAS); and in collaboration with the American College of Cardiology (ACC), American Heart Association (AHA), the Asia Pacific Heart Rhythm Society (APHRS), and the Society of Thoracic Surgeons (STS). Endorsed by the governing bodies of the American College of Cardiology Foundation, the American Heart Association, the European Cardiac Arrhythmia Society, the European Heart Rhythm Association, the Society of Thoracic Surgeons, the Asia Pacific Heart Rhythm Society, and the Heart Rhythm Society.
      • Gillinov A.M.
      • Gelijns A.C.
      • Parides M.K.
      • DeRose Jr., J.J.
      • Moskowitz A.J.
      • Voisine P.
      • et al.
      Surgical ablation of atrial fibrillation during mitral-valve surgery.
      The addition of a surgical ablation procedure does not increase morbidity or mortality and has even been associated with better perioperative outcomes.
      • Rankin J.S.
      • Grau-Sepulveda M.V.
      • Ad N.
      • Damiano Jr, R.J.
      • Gillinov A.M.
      • Brennan J.M.
      • et al.
      Associations between surgical ablation and operative mortality after mitral valve procedures.
      De novo AF, defined in this study as new AF 90 days or more after MV surgery (to exclude patients with only early postoperative AF), has been infrequently studied, and there are little data on incidence and risk factors for this.
      • Stulak J.M.
      • Suri R.M.
      • Dearani J.A.
      • Sundt III, T.M.
      • Schaff H.V.
      When should prophylactic maze procedure be considered in patients undergoing mitral valve surgery?.
      We therefore sought to (1) characterize the incidence over time of de novo AF in patients who underwent MV surgery with or without additional cardiac surgical procedures; (2) identify potentially modifiable preoperative or perioperative risk factors for de novo AF, such that a prophylactic AF ablation procedure might be considered; and (3) evaluate the sample size of a potential study designed to reduce the incidence of de novo AF (Video 1).
      Figure thumbnail fx2
      Video 1Dr Patrick M. McCarthy and Dr James L. Cox discuss aspects of the article. Video available at: https://www.jtcvs.org/article/S0022-5223(18)31035-3/fulltext.

      Materials and Methods

       Patients

      From April 2004 to December 2014, 2261 patients had mitral ± other cardiac procedures, and of these, 1370 had no AF history and therefore no surgical AF ablation. A preoperative Holter monitor was routinely obtained in any patient with symptoms suggestive of arrhythmias to confirm absence of AF. After excluding patients who refused to participate in the Bluhm Cardiovascular Institute's Clinical Trials Unit Cardiovascular Research Database (CARD) at Northwestern University (IRB#STU00012288), 1288 patients (94%) were included. Of those, 930 (72%) had follow-up heart rhythm information at 90 days or more after surgery and comprise the study cohort. Preoperative, intraoperative, and postoperative data were obtained from CARD, and the Society of Thoracic Surgery database definitions were used for complications. Reported stroke includes transient ischemic attack and cerebrovascular accident (hemorrhagic or ischemic). Postoperative AF includes patients with new AF or atrial flutter persisting longer than 1 hour or requiring treatment before discharge per Society of Thoracic Surgeons definitions.

       Atrial Fibrillation Monitoring and Clinical Follow-up Protocols

      Unlike Northwestern patients treated with AF ablation who undergo a regimen of postoperative AF monitoring,
      • Churyla A.
      • Iddriss A.
      • Andrei A.C.
      • Kruse J.
      • Malaisrie S.C.
      • Passman R.
      • et al.
      Biatrial or left atrial lesion set for ablation during mitral surgery: risks and benefits.
      these patients were not routinely monitored. Test results and office visits were prospectively collected by patient surveys at 6 and 12 months postdischarge and annually thereafter. Records pertaining to reported office visits, echocardiography and rhythm information, operations, cardioversions, catheter ablation, or hospitalizations were obtained and analyzed. Mortality data were aggregated continuously consulting sources that included (1) CARD registry; (2) reviews of medical records and correspondence with the treating physician; (3) online death indexes, including the Social Security Death Index and genealogy resources (ancestry.com); and (4) newspaper death notices. With the use of these sources, mortality follow-up information was available on 100% of the cohort.

       Statistical Analyses

      Data summaries included the mean and standard deviation or median and interquartile range for continuously distributed variables, such as age and left ventricular ejection faction. For discrete or categoric variables, such as gender or postsurgical complications, we have reported counts and percentages. Comparisons between patients who developed de novo AF during follow-up and those who did not were based on the 2-sample t test with unequal variances and Satterthwaite's approximation or the Wilcoxon rank-sum test for continuous variables. Comparisons involving discrete or categoric variables were based on the chi-square or Fisher exact test (when cell counts were <5).
      To estimate the probabilities of de novo AF, death, or AF-free survival, we used a 3-state Markov model that relies on the Aalen–Johansen estimator, as implemented in the R package etm.
      • Allignol A.
      • Schumacher M.
      • Beyersmann J.
      Empirical transition matrix of multi-state models: the etm package.
      Comparisons between patients with and without de novo AF were made at 1, 2, 3, 5, and 10 years after the index surgical procedure, both in the entire cohort and separately for those with degenerative mitral regurgitation (DMR), rheumatic, and ischemic MV cause, as well as the group with mixed causes or otherwise hard to classify that were grouped together as “other.” Overall survival was estimated using the Kaplan–Meier estimator.
      To identify univariable or multivariable risk factors for de novo AF during follow-up, the subdistribution functions were modeled using the Fine-Gray model implemented in the R package survival.
      • A package for survival analysis in S
      R package version 2.37-4 [computer program].
      We report (sub) hazard ratios and corresponding 95% confidence intervals (Table E1). To assess whether postoperative AF was associated with a higher risk of de novo AF during follow-up or with overall survival, we have used 1-to-1 propensity score (PS) matching. A total of 32 variables (included in the online Appendix, Table E2, and Figure E1), including demographics (age, body mass index), medical history, preoperative medications, and concomitant procedures (aortic valve [AV], TV surgery) were used to create the PS model of which the outcome was postoperative AF. A caliper of size 0.1 logit-PS standard deviation units was used. Association between TV surgery and risk of de novo AF was assessed using a 2-to-1 PS-matched analysis with a size 0.2 logit-PS standard deviation units caliper (Table E3 and Figure E2). Additional information is provided in the Appendix (Tables E4 and E5, Figures E3 and E4).
      By using methods for binomial proportions, we have estimated the required sample size for 2-arm randomized study designed to detect a 5-year 40% reduction of AF (from 13.6% to 8.2%) with 80% power at 2-sided 5% α-level. Because 406 of 930 participants (43.7%) were at risk at 5 years, we have applied an inflation factor of 930/406 = 2.02 to compensate for the loss to follow-up.
      Statistical significance was declared as the 2-sided 5% alpha level, and there were no adjustments for multiplicity. As such, on average, 1 in 20 associations identified might be spurious. All analyses were performed in R v. 3.1.1 (http://cran.r-project.org) and SAS v. 9.3 (SAS Institute, Inc, Cary, NC).

      Results

      The preoperative characteristics of study participants are summarized in Table 1. Concomitant procedures included AV surgery (n = 199; 21%), tricuspid valve (TV) surgery (n = 118; 13%), and coronary artery bypass grafting (n = 248; 27%). MV disease cause distribution was DMR (n = 479; 52%), rheumatic (n = 51; 5%), ischemic (n = 90; 10%), and “other” (n = 310; 33%). MV disease was treated by repair (n = 728; 78%) or replacement (n = 202; 22%). Rhythm assessment was based on 1 or more sources: electrocardiogram (881; 95%); Holter monitor (176; 19%); implantable pacemaker or loop recorder (130; 14%); outpatient telemetry (97; 10%); cardiac rehabilitation (59; 6%); or other (59; 6%). The mean follow-up time was 4.6 ± 3.0 years (median, 4.4; interquartile range, 1.9-7.0; maximum, 11.1 years).
      Table 1Baseline characteristics of patients by de novo atrial fibrillation status in follow-up
      VariableEntire cohort (N = 930)No AF in follow-up (N = 773)AF in follow-up (N = 157)P value
      Age59.9 ± 13.958.7 ± 13.865.4 ± 13.0<.001
      CHA2DS2-VASc2.3 ± 1.82.2 ± 1.82.9 ± 1.7<.001
      Ejection fraction60.0 (53.0, 65.0)60.0 (55.0, 65.0)60.0 (45.0, 63.0).004
      Left atrial size (cm) (N = 866)4.3 (3.8, 4.7)4.3 (3.8, 4.7)4.3 (3.8, 4.9).28
      Female392 (42)324 (42)68 (43).75
      Diabetes126 (14)97 (13)29 (18).048
      Hypertension518 (56)421 (54)97 (62).09
      Prior coronary artery bypass50 (5)36 (5)14 (9).031
      Prior pacemaker27 (3)17 (2)10 (6).005
      Congestive heart failure312 (34)247 (32)65 (41).022
      Coronary artery disease302 (34)233 (31)69 (45).001
      Repeat sternotomy105 (11)79 (10)26 (17).022
      Aortic stenosis103 (14)75 (12)28 (20).023
      Mitral stenosis92 (11)68 (10)24 (17).008
      New York Heart Association class.033
       Class I278 (30)244 (32)34 (22)
       Class II345 (38)280 (37)65 (42)
       Class III234 (25)185 (24)49 (31)
       Class IV63 (7)55 (7)8 (5)
      TR >2+165 (18)128 (17)37 (25).024
      MV cause<.001
       Degenerative479 (52%)424 (55%)55 (35%)
       Ischemic90 (10%)64 (8%)26 (17%)
       Rheumatic51 (5%)43 (6%)8 (5%)
       Other310 (33%)242 (31%)68 (43%)
      Values are mean ± standard deviation; n (%); or median (first quartile, third quartile). AF, Atrial fibrillation; CHA2DS2-VASc, congestive heart failure, hypertension, age, diabetes, stroke, vascular disease; TR, tricuspid regurgitation; MV, mitral valve.
      Of the 930 patients, 157 (17%) developed de novo AF. They were significantly older at the time of index surgery (65 ± 13 years vs 59 ± 14 years, P < .001), had more comorbidities, and underwent more complex surgeries (prior coronary artery bypass [9% vs 5%, P = .031], prior pacemaker implantation [6% vs 2%, P = .005], congestive heart failure [41% vs 32%, P = .022], and coronary artery disease [45% vs 31%, P = .001]) than patients who were AF-free during follow-up. In addition, patients with de novo AF underwent more concomitant procedures than those without de novo AF, including coronary artery bypass (37% vs 25%, P = .001), AV surgery (31% vs 19%, P = .001), and TV surgery (23% vs 11%, P < .001). The de novo AF group had significantly longer median perfusion and crossclamp times (Table 2), and the incidence of predischarge pacemaker implantation was higher (10% vs 3%, P < .001). Total median length of stay was longer (7 vs 6 days, P < .001), and the incidences of anticoagulant events (3% vs 0%, P = .001), antiarrhythmics on discharge (41% vs 23%, P < .001), and 30-day readmissions (17% vs 9%, P = .003) were higher in the de novo AF group. Patients who developed in-hospital postoperative AF were more likely to develop late de novo AF (28% vs 49%; P < .001) during follow-up.
      Table 2Operative, predischarge, and postoperative characteristics by de novo atrial fibrillation status in follow-up
      VariableEntire cohort (N = 930)No AF in follow-up (N = 773)AF in follow-up (N = 157)P value
      Perfusion time (min)98.5 (76.0, 135.0)96.0 (75.0, 134.0)115.0 (82.0, 152.0)<.001
      Crossclamp time (min)80.0 (63.0, 110.0)79.0 (62.0, 108.0)89.0 (68.0, 121.0).008
      Coronary artery bypass248 (27%)190 (25%)58 (37%).001
      AV surgery199 (21%)150 (19%)49 (31%).001
      TV surgery118 (13%)82 (11%)36 (23%)<.001
      Mitral surgery.012
       Repair728 (78%)617 (80%)111 (71%)
       Replacement202 (22%)156 (20%)46 (29%)
      Total intensive care unit, h31.0 (24.3, 68.8)30.5 (24.2, 66.9)44.3 (26.0, 72.3).038
      Total length of stay (d)6.0 (5.0, 9.0)6.0 (5.0, 9.0)7.0 (6.0, 11.0)<.001
      Predischarge pacemaker43 (5%)27 (3%)16 (10%)<.001
      Postoperative AF291 (31%)214 (28%)77 (49%)<.001
      Discharged on antiarrhythmics241 (26%)177 (23%)64 (41%)<.001
      Readmission within 30 d97 (10%)70 (9%)27 (17%).003
      Most recent postoperative TR (N = 879)<.001
       None/trivial383 (44%)330 (45%)53 (36%)
       Mild351 (40%)301 (41%)50 (34%)
       Moderate115 (13%)83 (11%)32 (22%)
       Moderate-severe13 (1%)9 (1%)4 (3%)
       Severe17 (2%)10 (1%)7 (5%)
      TR ≥3-4+ anytime in follow-up (N = 879)48 (5%)29 (4%)19 (13%)<.001
      Values are mean ± standard deviation; n (%); or median (first quartile, third quartile). AF, Atrial fibrillation; AV, aortic valve; TV, tricuspid valve; TR, tricuspid regurgitation.

       De Novo Atrial Fibrillation During Follow-up

      The incidence of de novo AF 1, 5, and 10 years after index surgery was 5.0%, 13.6%, and 22.7%, respectively (Figure 1, A). A trend toward an increased risk of stroke (Figure 2, A) was observed among patients who developed de novo AF (P = .055). The absolute risk of stroke was 5.8% for those who developed AF during follow-up and 1.7% among those who did not. The use of warfarin at last clinical follow-up was also significantly higher for patients with de novo AF (47.1% vs 14.5%; P < .001).
      Figure thumbnail gr1
      Figure 1Multi-state models summary of probabilities of de novo AF, death, and AF-free survival in (A) the entire cohort (N = 930) and (B) rheumatic MV cause. AF, Atrial fibrillation; MV, mitral valve.
      Figure thumbnail gr2
      Figure 2Freedom from stroke by AF status in follow-up (A) and overall survival by AF status in follow-up (B). MV, mitral valve; AF, Atrial fibrillation.
      Survival was not different (P = .41) among those who developed de novo AF during follow-up (Figure 2, B) and those who did not. Mortality varied depending on the underlying cause of MV disease. In the patients with DMR (Figure 3, A), mortality was 0%, 4.0%, and 7.9% at 1, 5, and 10 years, respectively. Among patients with ischemic mitral regurgitation (Figure 3, B), mortality at 10 years was 27.9%.
      Figure thumbnail gr3
      Figure 3Multi-state models summary of probabilities of de novo AF, death, and AF-free survival in (A) degenerative MV cause and (B) ischemic MV cause. AF, Atrial fibrillation.
      Among patients with rheumatic MV disease (Figure 1B), the incidence of de novo AF was 2% during the first year, increased to 15.8% at 5 years, and reached 23% at 10 years. Among patients with DMR (Figure 3, A), the incidence of de novo AF steadily increased (3.4% and 9.1% at 1 and 5 years, respectively) to 22.3% 10 years after index surgery. For patients with ischemic MV disease (Figure 3, B), the incidence of de novo AF was initially higher than the other causes (10.1% at 1 year, 15.1% at 5 years) but was similar by year 10 (22.9%). In unadjusted analyses, the risk of de novo AF was significantly higher in the ischemic MV disease group compared with the DMR group (hazard ratio, 2.31; 95% confidence interval, 1.45-3.69; P < .001) (Table 3). After covariate adjustments, this difference was no longer significant (hazard ratio, 1.48; 95% confidence interval, 0.90-2.42; P = .12) (Table 4).
      Table 3Summary of significant univariate risk factors for developing de novo atrial fibrillation
      VariableHR95% CIP value
      Age, y1.04(1.03-1.05)<.001
      Diabetes1.60(1.07-2.41).021
      Hypercholesterolemia1.43(1.04-1.96).027
      Hypertension1.56(1.13-2.16).007
      Prior cerebrovascular accident1.74(1.02-2.96).043
      Prior coronary artery bypass1.94(1.12-3.35).019
      Prior pacemaker3.10(1.63-5.90)<.001
      Congestive heart failure1.44(1.05-1.99).024
      Coronary artery disease1.79(1.30-2.46)<.001
      Repeat sternotomy1.84(1.21-2.81).005
      Aortic stenosis1.90(1.25-2.87).002
      Mitral stenosis1.70(1.10-2.65).018
      Coronary artery bypass1.66(1.20-2.30).003
      AV surgery1.89(1.35-2.65)<.001
      TV surgery2.34(1.62-3.40)<.001
      MV replacement (vs repair)1.74(1.24-2.46).002
      TV regurgitation
       None/trivialREFREFREF
       Mild1.33(0.92-1.94).133
       Moderate1.86(1.15-3.01).011
       Moderate/severe1.67(0.61-4.62).319
       Severe3.52(1.80-6.89)<.001
      TV regurgitation ≥2+1.87(1.29-2.72).001
      CHADS2 score
       0REFREFREF
       11.98(1.23-3.18).005
       22.78(1.69-4.57)<.001
       32.48(1.39-4.44).002
       43.80(1.77-8.15)<.001
       54.66(1.87-11.13)<.001
      MV cause
       DegenerativeREFREFREF
       Rheumatic1.44(0.68-3.02).34
       Ischemic2.31(1.45-3.69)<.001
       Other2.25(1.57-3.21)<.001
      HR, Hazard ratio; CI, confidence interval; AV, aortic valve; TV, tricuspid valve; MV, mitral valve; CHADS2, Congestive heart failure, hypertension, age, diabetes and stroke.
      Table 4Summary of multivariable risk factors for developing de novo atrial fibrillation in follow-up
      VariableHR95% CIP value
      Age1.03(1.02-1.05)<.001
      Prior pacemaker1.75(0.88-3.48).11
      Repeat sternotomy1.29(0.82-2.02).28
      AV surgery1.37(0.93-2.03).108
      TV surgery1.73(1.17-2.56).005
      MV cause
       DegenerativeREFREFREF
       Rheumatic0.98(0.45-2.16).96
       Ischemic1.48(0.90-2.42).12
       Other1.68(1.11-2.52).013
      HR, Hazard ratio; CI, confidence interval; AV, aortic valve; TV, tricuspid valve; MV, mitral valve.
      In PS-matched analyses, there was a trend toward an increased risk of de novo AF for patients who underwent TV surgery (P = .068) (Figure 4, A). After PS matching, patients who developed postoperative AF had a higher risk of de novo AF during follow-up (P = .015) (Figure 4, B).
      Figure thumbnail gr4
      Figure 4Estimated probabilities of developing de novo AF by (A) TV surgery status and (B) postoperative AF status in PS-matched groups. AF, Atrial fibrillation; TV, tricuspid valve.

       Tricuspid Regurgitation in Late Follow-up

      Patients who developed de novo AF had more late significant tricuspid regurgitation (TR) that was severe or greater (P < .001) (Table 2).

       Risk Factors for De Novo Atrial Fibrillation During Follow-up

      Preoperative and intraoperative univariable risk factors for developing de novo AF are displayed in Table 3. The multivariable model (Table 4) identified age at surgery (1.03 [1.02, 1.05]), TV surgery (1.73 [1.17, 2.56]), and “other” cause MR versus DMR (1.68 [1.11, 2.52]) as independent risk factors for developing de novo AF. Prior pacemaker (P = .11) and repeat sternotomy (P = .28) were not significantly associated with an increased de novo AF risk. Left atrial size and low ejection fraction were not found to be significant risk factors for de novo AF (Table E1).

      Discussion

      Summarization of AF in follow-up continues to remain challenging, despite several useful recent developments.
      • Andrei A.C.
      • McCarthy P.M.
      • Thomas J.D.
      • Abicht T.O.
      • Malaisrie S.C.
      • Li Z.
      • et al.
      Overcoming reporting challenges: How to display, summarize, and model late reintervention outcomes, follow-up, and vital status information after surgery for atrial fibrillation.
      • Li L.
      • Mao H.
      • Ishwaran H.
      • Rajeswaran J.
      • Ehrlinger J.
      • Blackstone E.H.
      Estimating the prevalence of atrial fibrillation from a three-class mixture model for repeated diagnoses.
      • Rajeswaran J.
      • Blackstone E.H.
      • Ehrlinger J.
      • Li L.
      • Ishwaran H.
      • Parides M.K.
      Probability of atrial fibrillation after ablation: using a parametric nonlinear temporal decomposition mixed effects model.
      • Ad N.
      • Holmes S.D.
      • Patel J.
      • Je H.G.
      • Shuman D.J.
      The need for consistent predictors of success for surgical ablation of atrial fibrillation: a call to action.
      Patients undergoing mitral surgery typically have atrial myopathy and dilated atria placing them at risk for AF. There are little data regarding how often this occurs in patients without preoperative AF. This study identified several risk factors for de novo AF, AF was not uncommon and increased over time, and there was a strong trend toward an association with late stroke. Although in our analyses, time to de novo AF development is measured since surgery time, findings remain virtually unchanged if time since hospital discharge is considered instead (Figure E10, Figure E11, Figure E12, Figure E5, Figure E6, Figure E7, Figure E8, Figure E9).
      A study from The Mayo Clinic investigated 573 patients with degenerative MR over 10 years and had similar findings to ours, with 11% and 23% de novo AF at 5 and 10 years compared with our finding of 9.1% and 22.3%, respectively.
      • Stulak J.M.
      • Suri R.M.
      • Dearani J.A.
      • Sundt III, T.M.
      • Schaff H.V.
      When should prophylactic maze procedure be considered in patients undergoing mitral valve surgery?.
      Risk factors for AF in that study were advanced age, diabetes, left atrial size greater than 50 mm, and more than mild TR, and as in our study, TV surgery did not increase the risk of late AF (although in our study there was a trend in the unadjusted analysis; P = .068) (Figure 4, A). However, Stulak and colleagues
      • Stulak J.M.
      • Suri R.M.
      • Dearani J.A.
      • Sundt III, T.M.
      • Schaff H.V.
      When should prophylactic maze procedure be considered in patients undergoing mitral valve surgery?.
      did find that AF reduced late survival (P = .001), which we did not. They did not report on late stroke rate.
      Unlike in the Mayo study, our cohort included patients with degenerative, rheumatic, ischemic, and other mixed-cause MV disease. The incidence of de novo AF remained similar among these 3 main groups at 10 years (degenerative 22.3%, rheumatic 23%, and ischemic 22.9%) despite mortality differences. Degenerative MV disease was associated with a steady rate of de novo AF and had the lowest mortality rate among the 3 groups at 10 years (7.1%). Patients with ischemic MV disease had a higher mortality at 10 years (27.9%) compared with the other groups, a reflection of the poor outcome associated with this disease. The group with other causes had a higher risk for de novo AF by multivariable analysis (Table 4).
      Among patients in whom de novo AF developed during follow-up, there was a trend toward an increased risk of stroke (P = .055). This risk potentially could be reduced with closure of the left atrial appendage at the time of surgery; however, this is not routinely performed in the absence of a surgical ablation procedure. If one is willing to assume a causal relationship between AF and late stroke, then prophylactic treatment of AF may reduce that risk. In our cohort, overall average follow-up was in fact significantly longer among patients who developed de novo AF in follow-up than for those who did not, making selection bias unlikely: 4.96 ± 2.90 years (no de novo AF) versus 5.98 ± 2.76 years (de novo AF) (P < .001). Alternative analytic approaches when selection bias is a potential concern include pattern mixture models for informative dropout.
      • Little R.J.
      Selection and pattern-mixture models.
      Our data show a univariate association between preoperative moderate or greater TR and the risk of de novo AF (25% if de novo AF, 17% if no de novo AF, P = .024). In addition, more patients with de novo AF had TR postoperatively. The question of whether TR leads to de novo AF or if de novo AF leads to worsening TR is an important distinction. In one study, right atrial dilation as a result of right heart volume overload in the setting of TR was shown to promote AF.
      • Stulak J.M.
      • Dearani J.A.
      • Puga F.J.
      • Zehr K.J.
      • Schaff H.V.
      • Danielson G.K.
      Right-sided Maze procedure for atrial tachyarrhythmias in congenital heart disease.
      Stulak and colleagues
      • Stulak J.M.
      • Schaff H.V.
      • Dearani J.A.
      • Orszulak T.A.
      • Daly R.C.
      • Sundt III, T.M.
      Restoration of sinus rhythm by the Maze procedure halts progression of tricuspid regurgitation after mitral surgery.
      demonstrated that AF after MV surgery increases the risk of progression to TR and that a successful Maze procedure at the time of surgery is associated with a decrease in risk. Some patients may benefit from prophylactic AF ablation to not only limit the development of de novo AF and stroke risk but also limit late TR.

       A Potential Trial of Prophylactic Atrial Fibrillation Ablation

      An adequately controlled randomized trial to detect a 5-year 40% reduction of AF with 80% power at a 2-sided 5% α-level would require 2392 patients. A targeted trial of prophylactic ablation AF for high-risk patients, however, such as those aged 65 years or more with multiple valve surgery, may be more feasible.

       Study Limitations

      We acknowledge several limitations of our retrospective study. First, our cohort is heterogeneous because it includes patients who underwent MV surgery for a spectrum of MV diseases, with concomitant cardiac procedures, such as coronary artery bypass grafting and other valvular operations. Second, patients in our cohort had no documented evidence of preoperative AF; however, it is possible that patients had undetected past episodes of AF. Third, we identified heart rhythm information mostly via electrocardiogram. We may have underreported de novo AF because patients were not followed per protocol, such as patients treated with AF ablation. We likely reported more on patients who developed symptomatic AF and sought treatment including rhythm assessment. Patients with de novo AF had a higher rate of pacemaker implant predischarge that allowed a more thorough rhythm assessment. Finally, we did not have the clinical information to determine whether AF was paroxysmal, persistent, or long-standing persistent.

      Conclusions

      A steady rate of de novo AF (>90 days postsurgery) develops after MV surgery in patients without a known preoperative history of AF. The development of de novo AF was associated with a trend toward increased stroke risk, but not an overall survival difference. A trial of prophylactic AF ablation, especially in high-risk patients, to limit the development of late AF, stroke, and late TR may be warranted.

       Conflict of Interest Statement

      A.-C.A.: Atricure consultant. J.L.C.: Atricure consultant, Chairman, Scientific Advisory Board, stockholder. All other authors have nothing to disclose with regard to commercial support.

      Appendix

       1. Evaluation of Potential Risk Factors for Developing De Novo AF in Follow-up

      In addition to the variable presented in the article's main body, we have explored several other potential risk factors for developing de novo AF. Results are shown in the following table.

       2. Additional Details Regarding the Propensity Score–Matched Analyses

       Postoperative atrial fibrillation versus no postoperative atrial fibrillation comparison

      The PS model has modeled the probability of developing postoperative AF based on a logistic regression model with the following covariates: age at the time of surgery, gender, body mass index, creatinine level, left ventricular ejection fraction, diabetes, hypercholesterolemia, hypertension, chronic obstructive pulmonary disease, prior myocardial infarction, history of congestive heart failure, prior coronary artery bypass, prior heart valve procedure, repeat sternotomy, prior cardiovascular intervention, cerebrovascular disease, peripheral vascular disease, prior stroke, prior pacemaker, elective surgery, AV surgery, MV repair or replacement, coronary artery bypass grafting, New York Heart Association Functional Class III or IV, and preoperative medication (beta-blockers, anticoagulants, aspirin, statins, and diuretics).
      We have modeled the probability of experiencing postoperative AF and carried out 1-to-1 PS matching with a caliper of size 0.1 logit-PS standard deviation units. The area under the receiver's operating characteristics curve was equal to 0.713, and the Hosmer–Lemeshow test was chi-square = 6.67, degrees of freedom = 8, P value = .57. Summaries of baseline and intraoperative characteristics after PS matching by postoperative AF status are shown in Table E2. Corresponding standardized mean differences are depicted in Figure E1. They confirm that adequate covariate balance has been achieved after PS matching.

       TV surgery versus no TV surgery comparison

      A similar PS-matched analysis was performed using the same variables as predictors. The caliper PS-matching process used a 2-to-1 matching ratio (no TV surgery vs TV surgery) and was based on a Greedy algorithm. A caliper of size 0.1 logit-PS standard deviation units was used. The area under the receiver's operating characteristics curve was equal to 0.753. The Hosmer–Lemeshow test of calibration was chi-square = 6.62, degrees of freedom = 8, P = .58. Summaries of baseline and intraoperative characteristics before and after PS matching by TV surgery status are shown in Table E3. Corresponding standardized mean differences are depicted in Figure E2. They confirm that adequate covariate balance has been achieved after PS matching.
      Figure thumbnail fx3
      Figure E1Standardized mean differences before and after PS matching by postoperative AF status. PS, Propensity score; CHADS2, congestive heart failure, hypertension, age, diabetes, stroke; CHADS2VASC, congestive heart failure, hypertension, age, diabetes, stroke, vascular disease; MI, myocardial infarction; CABG, coronary artery bypass grafting; NYHA, New York Heart Association; TR, tricuspid regurgitation; AV, aortic valve; MV, mitral valve.
      Figure thumbnail fx4
      Figure E2Standardized mean differences before and after PS matching by TV surgery status. PS, Propensity score; CHADS2, congestive heart failure, hypertension, age, diabetes, stroke; CHADS2VASC, congestive heart failure, hypertension, age, diabetes, stroke, vascular disease; MI, myocardial infarction; CABG, coronary artery bypass grafting; NYHA, New York Heart Association; AV, aortic valve; MV, mitral valve.
      Figure thumbnail fx5
      Figure E3Histogram of the number of AF assessment (diagnosis) times during the postsurgical follow-up of the entire cohort. AF, Atrial fibrillation.
      Figure thumbnail fx6
      Figure E4Estimates of the de novo AF incidence based on methods for interval-censored data. AF, Atrial fibrillation; MV, mitral valve.
      Figure thumbnail fx7
      Figure E5Multi-state models summary of probabilities of de novo AF, death, and AF-free survival in the entire cohort. AF, Atrial fibrillation.
      Figure thumbnail fx8
      Figure E6Multi-state models summary of probabilities of de novo AF, death, and AF-free survival in rheumatic MV cause. AF, Atrial fibrillation.
      Figure thumbnail fx9
      Figure E7Multi-state models summary of probabilities of de novo AF, death, and AF-free survival in degenerative MV cause. AF, Atrial fibrillation.
      Figure thumbnail fx10
      Figure E8Multi-state models summary of probabilities of de novo AF, death, and AF-free survival in ischemic MV cause. AF, Atrial fibrillation.
      Figure thumbnail fx11
      Figure E9Freedom from stroke by AF status in follow-up. AF, Atrial fibrillation.
      Figure thumbnail fx12
      Figure E10Overall survival by AF status in follow-up. AF, Atrial fibrillation.
      Figure thumbnail fx13
      Figure E11Estimated probabilities of developing de novo AF by TV surgery status in PS-matched groups. AF, Atrial fibrillation; TV, tricuspid valve.
      Figure thumbnail fx14
      Figure E12Estimated probabilities of developing de novo AF by postoperative AF status in PS-matched groups. AF, Atrial fibrillation.
      Table E1Summary of left atrial diameter, left ventricular ejection fraction, history of congestive heart failure, and mitral valve cause as potential univariate risk factors for developing de novo atrial fibrillation
      VariableHR95% CIP value
      Left atrial diameter (continuous)1.16(0.95-1.43).146
      Left atrial diameter (categoric)
       <5 cmREFREFREF
       5-6 cm1.44(0.96-2.17).08
       ≥6 cm1.30(0.41-4.10).65
      Left ventricular ejection fraction (continuous)1.01(0.98-1.002).095
      Left ventricular ejection fraction (categoric)
       >45%REFREFREF
       30%-45%2.11(1.42-3.13)<.001
       ≤30%1.04(0.60-1.82).88
      History of congestive heart failure1.44(1.05-1.99).024
      MV cause
       DegenerativeREFREFREF
       Rheumatic1.44(0.68-3.02).34
       Ischemic2.31(1.45-3.69)<.001
       Other2.25(1.57-3.21)<.001
      HR, Hazard ratio; CI, confidence interval; MV, mitral valve.
      Table E2Summary of baseline characteristics in propensity score–matched groups by postoperative atrial fibrillation status
      VariableNo postoperative AF (N = 259)Postoperative AF

      (N = 259)
      P value
      Age (y)64.7 ± 11.964.4 ± 11.3.76
      Body mass index (kg/m2)27.5 ± 6.027.3 ± 5.3.69
      CHADS21.5 ± 1.21.5 ± 1.2.91
      CHADS2VASC2.6 ± 1.72.5 ± 1.7.55
      Creatinine level, median (Q1, Q3)1.0 (0.8, 1.1)1.0 (0.9, 1.2).35
      Ejection fraction, median (Q1, Q3)60.0 (50.0, 65.0)60.0 (50.0, 65.0).84
      Left atrial size, median (Q1, Q3)4.3 (3.7, 4.7)4.3 (3.8, 4.8).37
      Gender (female)110 (42%)103 (40%).53
      Diabetes40 (15%)40 (15%)1.00
      Dyslipidemia140 (54%)145 (56%).66
      Hypertension172 (66%)166 (64%).58
      Chronic lung disease36 (14%)32 (12%).60
      Peripheral vascular disease15 (6%)12 (5%).55
      Cerebrovascular disease21 (8%)20 (8%).87
      Prior stroke14 (5%)16 (6%).71
      Previous myocardial infarction32 (12%)30 (12%).79
      Prior CABG14 (5%)17 (7%).58
      Prior pacemaker5 (2%)3 (1%).48
      Congestive heart failure88 (34%)91 (35%).78
      NYHA class III or IV87 (34%)90 (35%).78
      Coronary artery disease98 (38%)97 (38%).98
      Repeat sternotomy21 (8%)28 (11%).29
      Elective surgery220 (85%)221 (85%).90
      AV stenosis34 (16%)36 (17%).88
      MV stenosis26 (11%)31 (13%).52
      Preoperative anticoagulants18 (7%)20 (8%).74
      Tricuspid regurgitation >2+47 (19%)60 (24%).14
      AV insufficiency.10
       0 = None/trivial141 (65%)154 (69%)
       1 = Mild50 (23%)31 (14%)
       2 = Moderate15 (7%)23 (10%)
       3 = Moderate/severe1 (0%)3 (1%)
       4 = Severe10 (5%)12 (5%)
      MV insufficiency.62
       0 = None/trivial3 (1%)7 (3%)
       1 = Mild18 (7%)17 (7%)
       2 = Moderate39 (15%)46 (18%)
       3 = Moderate/severe23 (9%)23 (9%)
       4 = Severe176 (68%)164 (64%)
      Coronary artery bypass85 (33%)77 (30%).45
      AV surgery61 (24%)63 (24%).84
      TV surgery39 (15%)38 (15%).90
      MV surgery type.92
       Repair194 (75%)195 (75%)
       Replacement65 (25%)64 (25%)
      AF, Atrial fibrillation; CHADS2, congestive heart failure, hypertension, age, diabetes, stroke; CHADS2VASC, congestive heart failure, hypertension, age, diabetes, stroke, vascular disease; CABG, coronary artery bypass grafting; NYHA, New York Heart Association; AV, aortic valve; MV, mitral valve; TV, tricuspid valve.
      Table E3Summary of baseline characteristics in propensity score–matched groups by tricuspid valve surgery status
      VariableNo TV surgery (N = 208)TV surgery (N = 110)P value
      Age63.2 ± 14.263.4 ± 13.7.90
      Body mass index (kg/m2)26.4 ± 5.526.4 ± 5.81.00
      CHADS21.7 ± 1.41.7 ± 1.2.96
      CHADS2VASC3.0 ± 2.03.0 ± 1.6.97
      Creatinine level, median (Q1, Q3)1.0 (0.8, 1.2)1.0 (0.9, 1.3).49
      Ejection fraction, median (Q1, Q3)60.0 (44.5, 65.0)60.0 (45.0, 65.0).45
      Left atrial size, median (Q1, Q3)4.3 (3.8, 4.7)4.3 (3.8, 4.9).28
      Gender (female)112 (54%)60 (55%).91
      Diabetes47 (23%)24 (22%).87
      Dyslipidemia108 (52%)56 (51%).86
      Hypertension115 (55%)61 (55%).98
      Chronic lung disease24 (12%)11 (10%).68
      Peripheral vascular disease13 (6%)6 (5%).78
      Cerebrovascular disease23 (11%)11 (10%).77
      Prior stroke15 (7%)8 (7%).98
      Previous myocardial infarction38 (18%)17 (15%).53
      Prior coronary artery bypass graft23 (11%)12 (11%).97
      Prior pacemaker8 (4%)7 (6%).31
      Congestive heart failure103 (50%)54 (49%).94
      NYHA class III or IV95 (46%)50 (45%).97
      Coronary artery disease89 (45%)44 (41%).54
      Repeat sternotomy41 (20%)20 (18%).74
      Elective surgery166 (80%)87 (79%).88
      AV stenosis24 (13%)19 (21%).12
      MV stenosis32 (16%)19 (19%).55
      Preoperative anticoagulants23 (11%)14 (13%).66
      TR >2+24 (12%)81 (80%)<.001
      AV insufficiency.70
       0 = None/trivial120 (66%)60 (62%)
       1 = Mild35 (19%)19 (20%)
       2 = Moderate15 (8%)8 (8%)
       3 = Moderate/severe3 (2%)1 (1%)
       4 = Severe9 (5%)9 (9%)
      MV insufficiency.92
       0 = None/trivial9 (4%)4 (4%)
       1 = Mild16 (8%)9 (8%)
       2 = Moderate37 (18%)17 (15%)
       3 = Moderate/severe17 (8%)12 (11%)
       4 = Severe128 (62%)68 (62%)
      Coronary artery bypass grafting71 (34%)34 (31%).56
      AV surgery42 (20%)24 (22%).73
      MV surgery type.74
       Repair140 (67%)72 (65%)
       Replacement68 (33%)38 (35%)
      TV, Tricuspid valve; CHADS2, congestive heart failure, hypertension, age, diabetes, stroke; CHADS2VASC, congestive heart failure, hypertension, age, diabetes, stroke, vascular disease; NYHA, New York Heart Association; AV, aortic valve; MV, mitral valve; TR, tricuspid regurgitation.
      Table E4Summary of significant univariate risk factors for developing de novo atrial fibrillation
      VariableModels for time since surgery dateModels for time since hospital discharge date
      HR (95% CI)P valueHR (95% CI)P value
      Age, y1.04 (1.03-1.05)<.0011.04 (1.03-1.05)<.001
      Diabetes1.60 (1.07-2.41).0211.61 (1.08-2.41).028
      Hypercholesterolemia1.43 (1.04-1.96).0271.43 (1.04-1.96).027
      Hypertension1.56 (1.13-2.16).0071.56 (1.13-2.16).006
      Prior cerebrovascular accident1.74 (1.02-2.96).0431.74 (1.03-2.98).056
      Prior coronary artery bypass1.94 (1.12-3.35).0191.94 (1.12-3.36).031
      Prior pacemaker3.10 (1.63-5.90)<.0013.10 (1.63-5.91).003
      Congestive heart failure1.44 (1.05-1.99).0241.45 (1.05-1.99).025
      Coronary artery disease1.79 (1.30-2.46)<.0011.79 (1.30-2.46)<.001
      Repeat sternotomy1.84 (1.21-2.81).0051.85 (1.21-2.82).008
      Aortic stenosis1.90 (1.25-2.87).0021.90 (1.26-2.88).005
      Mitral stenosis1.70 (1.10-2.65).0181.71 (1.10-2.65).025
      Coronary artery bypass1.66 (1.20-2.30).0031.66 (1.20-2.30).003
      AV surgery1.89 (1.35-2.65)<.0011.89 (1.35-2.66)<.001
      TV surgery2.34 (1.62-3.40)<.0012.35 (1.62-3.41)<.001
      MV replacement (vs repair)1.74 (1.24-2.46).0021.75 (1.24-2.47).002
      TV regurgitation
       None/trivialREFREFREFREF
       Mild1.33 (0.92-1.94).1331.33 (0.91-1.94).135
       Moderate1.86 (1.15-3.01).0111.86 (1.15-3.01).012
       Moderate/severe1.67 (0.61-4.62).3191.67 (0.61-4.61).321
       Severe3.52 (1.80-6.89)<.0013.55 (1.82-6.95)<.001
       TV regurgitation ≥2+1.87 (1.29-2.72).0011.87 (1.29-2.72).002
      CHADS2 score
       0REFREFREFREF
       11.98 (1.23-3.18).0051.99 (1.24-3.19).004
       22.78 (1.69-4.57)<.0012.78 (1.69-4.57)<.001
       32.48 (1.39-4.44).0022.49 (1.40-4.46).002
       43.80 (1.77-8.15)<.0013.83 (1.78-8.21)<.001
       54.66 (1.87-11.13)<.0014.57 (1.87-11.17)<.001
      MV cause
       DegenerativeREFREFREFREF
       Rheumatic1.44 (0.68-3.02).341.44 (0.68-3.02).34
       Ischemic2.31 (1.45-3.69)<.0012.32 (1.45-3.69)<.001
       Other2.25 (1.57-3.21)<.0012.26 (1.58-3.23)<.001
      HR, Hazard ratio; CI, confidence interval; AV, aortic valve; TV, tricuspid valve; MV, mitral valve; CHADS2, congestive heart failure, hypertension, age, diabetes and stroke.
      Table E5Summary of multivariable risk factors for developing de novo atrial fibrillation in follow-up
      VariableModels for time since surgery dateModels for time since hospital discharge date
      HR (95% CI)P valueHR (95% CI)P value
      Age1.03 (1.02-1.05)<.0011.03 (1.02-1.05)<.001
      Prior pacemaker1.75 (0.88-3.48).111.74 (0.88-3.47).11
      Repeat sternotomy1.29 (0.82-2.02).281.29 (0.82-2.03).27
      AV surgery1.37 (0.93-2.03).1081.37 (0.93-2.03).109
      TV surgery1.73 (1.17-2.56).0051.74 (1.18-2.57).005
      MV cause
       DegenerativeREFREFREFREF
       Rheumatic0.98 (0.45-2.16).960.98 (0.45-2.16).97
       Ischemic1.48 (0.90-2.42).121.48 (0.91-2.43).12
       Other1.68 (1.11-2.52).0131.68 (1.12-2.53).013
      HR, Hazard ratio; CI, confidence interval; AV, aortic valve; TV, tricuspid valve; MV, mitral valve.

      Supplementary Data

      References

        • Lim E.
        • Barlow C.W.
        • Hosseinpour A.R.
        • Wisbey C.
        • Wilson K.
        • Pidgeon W.
        • et al.
        Influence of atrial fibrillation on outcome following mitral valve repair.
        Circulation. 2001; 104: I59-I63
        • Ngaage D.L.
        • Schaff H.V.
        • Mullany C.J.
        • Barnes S.
        • Dearani J.A.
        • Daly R.C.
        • et al.
        Influence of preoperative atrial fibrillation on late results of mitral repair: is concomitant ablation justified?.
        Ann Thorac Surg. 2007; 84: 434-443
        • Calkins H.
        • Kuck K.H.
        • Cappato R.
        • Brugada J.
        • Camm A.J.
        • Chen S.A.
        • 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. Developed in partnership with the European Heart Rhythm Association (EHRA), a registered branch of the European Society of Cardiology (ESC) and the European Cardiac Arrhythmia Society (ECAS); and in collaboration with the American College of Cardiology (ACC), American Heart Association (AHA), the Asia Pacific Heart Rhythm Society (APHRS), and the Society of Thoracic Surgeons (STS). Endorsed by the governing bodies of the American College of Cardiology Foundation, the American Heart Association, the European Cardiac Arrhythmia Society, the European Heart Rhythm Association, the Society of Thoracic Surgeons, the Asia Pacific Heart Rhythm Society, and the Heart Rhythm Society.
        Heart Rhythm. 2012; 9: 632-696.e621
        • Badhwar V.
        • Rankin J.S.
        • Damiano Jr., R.J.
        • Gillinov A.M.
        • Bakaeen F.G.
        • Edgerton J.R.
        • et al.
        The Society of Thoracic Surgeons 2017 clinical practice guidelines for the surgical treatment of atrial fibrillation.
        Ann Thorac Surg. 2017; 103: 329-341
        • Gillinov A.M.
        • Gelijns A.C.
        • Parides M.K.
        • DeRose Jr., J.J.
        • Moskowitz A.J.
        • Voisine P.
        • et al.
        Surgical ablation of atrial fibrillation during mitral-valve surgery.
        N Engl J Med. 2015; 372: 1399-1409
        • Rankin J.S.
        • Grau-Sepulveda M.V.
        • Ad N.
        • Damiano Jr, R.J.
        • Gillinov A.M.
        • Brennan J.M.
        • et al.
        Associations between surgical ablation and operative mortality after mitral valve procedures.
        Ann Thorac Surg. 2018; 105: 1790-1796
        • Stulak J.M.
        • Suri R.M.
        • Dearani J.A.
        • Sundt III, T.M.
        • Schaff H.V.
        When should prophylactic maze procedure be considered in patients undergoing mitral valve surgery?.
        Ann Thorac Surg. 2010; 89: 1395-1401
        • Churyla A.
        • Iddriss A.
        • Andrei A.C.
        • Kruse J.
        • Malaisrie S.C.
        • Passman R.
        • et al.
        Biatrial or left atrial lesion set for ablation during mitral surgery: risks and benefits.
        Ann Thorac Surg. 2017; 103: 1858-1865
        • Allignol A.
        • Schumacher M.
        • Beyersmann J.
        Empirical transition matrix of multi-state models: the etm package.
        J Stat Softw. 2011; 38: 1-15
        • A package for survival analysis in S
        R package version 2.37-4 [computer program].
        (Available at:)
        • Andrei A.C.
        • McCarthy P.M.
        • Thomas J.D.
        • Abicht T.O.
        • Malaisrie S.C.
        • Li Z.
        • et al.
        Overcoming reporting challenges: How to display, summarize, and model late reintervention outcomes, follow-up, and vital status information after surgery for atrial fibrillation.
        Heart Rhythm. 2015; 12: 1456-1463
        • Li L.
        • Mao H.
        • Ishwaran H.
        • Rajeswaran J.
        • Ehrlinger J.
        • Blackstone E.H.
        Estimating the prevalence of atrial fibrillation from a three-class mixture model for repeated diagnoses.
        Biom J. 2017; 59: 331-343
        • Rajeswaran J.
        • Blackstone E.H.
        • Ehrlinger J.
        • Li L.
        • Ishwaran H.
        • Parides M.K.
        Probability of atrial fibrillation after ablation: using a parametric nonlinear temporal decomposition mixed effects model.
        Stat Methods Med Res. 2018; 27: 126-141
        • Ad N.
        • Holmes S.D.
        • Patel J.
        • Je H.G.
        • Shuman D.J.
        The need for consistent predictors of success for surgical ablation of atrial fibrillation: a call to action.
        Innovations. 2017; 12: 421-429
        • Little R.J.
        Selection and pattern-mixture models.
        in: Fitzmaurice G. Davidian M. Verbeke G. Molenberghs G. Advances in Longitudinal Data Analysis. CRC Press, London2008: 409-431
        • Stulak J.M.
        • Dearani J.A.
        • Puga F.J.
        • Zehr K.J.
        • Schaff H.V.
        • Danielson G.K.
        Right-sided Maze procedure for atrial tachyarrhythmias in congenital heart disease.
        Ann Thorac Surg. 2006; 81: 1780-1785
        • Stulak J.M.
        • Schaff H.V.
        • Dearani J.A.
        • Orszulak T.A.
        • Daly R.C.
        • Sundt III, T.M.
        Restoration of sinus rhythm by the Maze procedure halts progression of tricuspid regurgitation after mitral surgery.
        Ann Thorac Surg. 2008; 86: 40-45