- Email: [email protected]
Burden of preoperative atrial fibrillation in patients undergoing coronary artery bypass grafting
Malaisrie et al
Acquired
Burden of preoperative atrial fibrillation in patients undergoing coronary artery bypass grafting S. Chris Malaisrie, MD,a Patrick M. McCarthy, MD,a Jane Kruse, BSN,a Roland Matsouaka, PhD,b Adin-Cristian Andrei, PhD,a Maria V. Grau-Sepulveda, MD,b Daniel J. Friedman, MD,b James L. Cox, MD,a and J. Matthew Brennan, MDb ABSTRACT Background: This study compares early and late outcomes in patients undergoing coronary artery bypass grafting with and without preoperative atrial fibrillation in a contemporary, nationally representative Medicare cohort.
Results: Preoperative atrial fibrillation was associated with a higher adjusted 30day mortality (odds ratio [OR], 1.5; P <.0001) and combined major morbidity including stroke, renal failure, prolonged ventilation, reoperation, and deep sternal wound infection (OR, 1.32; P<.0001). Patients with preoperative atrial fibrillation experienced a higher adjusted long-term risk of all-cause mortality and cumulative risk of stroke and systemic embolism compared to those without atrial fibrillation. At 5 years, the survival probability in the preoperative atrial fibrillation versus no atrial fibrillation groups stratified by CHA2DS2-VASc scores was 74.8% versus 86.3% (score 1-3), 56.5% versus 73.2% (score 4-6), and 41.2% versus 57.2% (score 7-9; all P <.001).
Survival is worse in patients with preoperative AF compared to those with no AF (P < .001). Central Message AF before CABG is associated with higher early and late perioperative mortality and morbidity. Survival and risk of stroke or systemic embolism are worse after adjusting for comorbidities.
!ACQ
Methods: In the Medicare-Linked Society of Thoracic Surgeons database, 361,138 patients underwent isolated coronary artery bypass from 2006 to 2013, of whom 37,220 (10.3%) had preoperative atrial fibrillation; 13,161 (35.4%) were treated with surgical ablation and were excluded. Generalized estimating equations were used to compare 30-day mortality and morbidity. Long-term survival was summarized using Kaplan-Meier curves and Cox regression models. Stroke and systemic embolism incidence was modeled using the Fine-Gray model and the CHA2DS2-VASc score was used to analyze stroke risk. Median follow-up was 4 years.
Perspective Preoperative atrial fibrillation is an independent risk factor for worse outcomes after coronary artery bypass surgery. Further analysis of the comparative effectiveness of concomitant atrial fibrillation ablation has important implications for this high-risk cohort.
See Editorial Commentary page XXX.
Conclusions: Preoperative atrial fibrillation is independently associated with worse early and late postoperative outcomes. CHA2DS2-VASc stratifies risk, even in those without preoperative atrial fibrillation. (J Thorac Cardiovasc Surg 2018;-:1-10)
The incidence of atrial fibrillation in patients with coronary artery disease (CAD) requiring coronary artery bypass From the aDivision of Cardiac Surgery, Northwestern University, Bluhm Cardiovascular Institute, Chicago, Ill; and bDuke Clinical Research Institute, Duke University School of Medicine, Durham, NC. This work was supported by institutional funding from Northwestern University. Read at the 43rd Annual Meeting of The Western Thoracic Surgical Association, Colorado Springs, Colorado, June 21-24, 2017. Received for publication June 21, 2017; revisions received Dec 5, 2017; accepted for publication Jan 9, 2018. Address for reprints: S. Chris Malaisrie, MD, Division of Cardiac Surgery at Northwestern University Feinberg School of Medicine and Northwestern, Bluhm Cardiovascular Institute, 201 East Huron St, Suite 11-140, Chicago, IL 60611-2908 (E-mail: [email protected]). 0022-5223/$36.00 Copyright Ó 2018 by The American Association for Thoracic Surgery https://doi.org/10.1016/j.jtcvs.2018.01.069
grafting (CABG) is approximately 6% according to a historical Society of Thoracic Surgeons (STS) report.1 Published reports on the association between preoperative atrial fibrillation (AF) and post-CABG outcomes have shown worse perioperative mortality and long-term survival.2-8 Previous STS database studies have shown increased perioperative mortality and stroke in patients with preoperative AF, but long-term outcomes were not studied.9 Scanning this QR code will take you to a supplemental figure and video for the article.
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
1
Acquired
!ACQ
Abbreviations and Acronyms ASCERT ¼ American College of Cardiology Foundation–Society of Thoracic Surgeons Collaboration on the Comparative Effectiveness of Revascularization Strategies AF ¼ atrial fibrillation CABG ¼ coronary artery bypass graft CAD ¼ coronary artery disease CHA2DS2-VASc ¼ Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, VAScular disease, Age 65-74, Sex category CI ¼ confidence interval CMS ¼ Centers for Medicare and Medicaid HR ¼ hazard ratio OR ¼ odds ratio SSE ¼ stroke or systemic embolism STS ¼ Society of Thoracic Surgeons
The STS database recent linkage to the Centers for Medicare and Medicaid Services (CMS) database allows for study of important long-term endpoints beyond perioperative outcomes provided in the STS National Cardiac Database. Using CMS-linked STS data, this study sought to examine early and late outcomes in 2 groups of first cardiac surgery CABG patients: (1) those with preoperative AF but no surgical AF treatment, and (2) those without preoperative AF. PATIENTS AND METHODS Data for all patients were obtained from the STS Adult Cardiac Surgery Database (versions 2.52, 2.61, and 2.73) for patients discharged between January 1, 2006, and December 31, 2013, and who could be linked to CMS data using a validated deterministic matching algorithm.10
Study Population Medicare recipients age 65 years or older and undergoing CABG as a first cardiac surgery who had complete information on preoperative history of atrial arrhythmias were included. Between January 2006 and December 2013, there were 688,466 isolated CABG surgery procedures reported in the STS database in patients age 65 years or older. Of those, 119,597 patients were excluded because of prior cardiac surgery (n ¼ 44,671), emergent/salvage status–cardiogenic shock/resuscitation (n ¼ 25,542), preoperative intra-aortic balloon pump/inotropes (n ¼ 44,977), history of endocarditis (n ¼ 637), and missing or inconsistent data on arrhythmia, number of diseased vessels, gender, or any of the previous fields considered for exclusion (n ¼ 3770). Of the resulting population, additional exclusions were made for 207,553 with unlinked CMS data, 178 had more than 1 admission on the same day, and 13,161 had surgical ablation. The resulting 347,977 patients with linked CMS data (61%) were categorized according to their preoperative AF status: (1) 89.7% (n ¼ 323,918) no preoperative
2
Malaisrie et al
AF (No AF) or (2) 6.7% (n ¼ 24,059) preoperative AF (AF group) and analyzed (Figure E1).
Study Outcomes The primary outcome was all-cause mortality after CABG. This was defined using STS registry data for in-hospital deaths and the linked Medicare Denominator File for postdischarge deaths.11 Secondary outcomes included (1) stroke or systemic embolism (SSE; ischemic stroke, hemorrhagic stroke, transient ischemic attack, or systemic arterial embolism), (2) in-hospital mortality, and (3) a composite of inhospital major morbidity.12 Incident SSE was defined by using STS registry data to account for inhospital strokes and Medicare Part A data to identify subsequent rehospitalizations with SSE as a primary diagnosis (International Classification of Diseases, 9th revision, codes: 433.1, 434.x1, 430, 431, 432.0, 432.1, 432.9, 444,x, 435.x). The STS database does not include the exact date of in-hospital strokes; for the purpose of time-to-event analyses, we assigned the date of in-hospital stroke to be the date of the index operation. In-hospital major morbidity, a previously defined composite,13,14 referred to any of the following postprocedure complications: permanent stroke, new cases of renal failure, prolonged ventilation (ventilation longer than 24 hours after surgery), reoperation for cardiac reasons (graft dysfunction, bleeding, valve dysfunction or other), and deep sternal wound infection.
Statistical Methods The overall study population and the subgroups of interest (No AF vs AF) were summarized using medians and interquartile ranges (25th and 75th percentiles) for continuous variables and frequency counts and percentages for categorical variables. Differences in distributions among patients with No AF versus AF were evaluated with Wilcoxon and Pearson chi-square tests, respectively. Logistic regression models were developed to estimate the association between in-hospital outcomes and AF versus No AF. The generalized estimated equations method with an exchangeable correlation structure was used to compute confidence limits that account for hospital clustering of patients. Adjusted models included predictors of long-term survival after CABG identified with the validated American College of Cardiology Foundation–Society of Thoracic Surgeons Collaboration on the Comparative Effectiveness of Revascularization Strategies (ASCERT) model15: age, ejection fraction, year of surgery, renal failure (glomerular filtration rate <30 or dialysis), glomerular filtration rate, body mass index, height, sex, race, current smoking, chronic heart failure/New York Heart Association Class IV, urgent operative status, number of diseased vessels, diabetes, previous carotid surgery, carotid stenosis > 75%, transient ischemic attack, chronic lung disease moderate/severe, left main disease, unstable angina, peripheral vascular disease, immunosuppressant medication, hypertension, valve insufficiency (tricuspid, aortic, mitral) moderate/severe, aortic stenosis gradient, preoperative myocardial infarction (less than 6 hours before, between 6 and 24 hours, between 1 and 21 days). These variables were also used for all adjusted time-to-event analyses. Time-to-event analysis was used to compare long-term survival and SSE occurrence by group. For survival, patient follow-up was censored at the end of study period (January 1, 2014). Product-limit Kaplan Meier survival estimates were computed for each group and comparisons were made with log-rank tests. Cox regression models were used to compute hazard ratios for the AF versus No AF, both unadjusted and adjusted by other survival predictors. A robust sandwich covariance estimation to compute 95% confidence intervals was used to account for correlation of patients’ failure times within the same hospital. The proportional hazard assumption was tested using log-log survival plots (log-log survival vs log-time) for each group and the results indicated no violation of this assumption. Consistent with the validated STS risk models,15 missing values (<3%) were imputed with relevant groups-specific medians for
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al
Acquired
TABLE 1. Baseline characteristics of patients in No AF or AF groups No AF (N ¼ 323,918)
Variable* Age, y
72 (68, 77)
Female sex 2
28.09 (25, 32)
Height, cm
172.7 (163, 178)
75 (71, 80) 6920 (29) 28.10 (25, 32) 173 (165, 180)
Standardized difference
P value
39.94
<.0001
4.17
<.0001
1.93
.954
12.06
<.0001 <.0001
Race Other Asian Hispanic Black Caucasian Missing Current smoking
6725 (2.08) 5138 (1.59) 11,218 (3.46) 16,326 (5.04) 283,245 (87.44) 1266 (0.39)
428 (1.78) 283 (1.18) 548 (2.28) 751 (3.12) 21,952 (91.24) 97 (0.40)
2.16 3.51 7.10 9.71 12.33 0.20
46,294 (14.29)
3111 (12.93)
4.00
<.0001
Diabetes
129,348 (40)
9903 (41)
2.50
<.0001
Hypertension
283,689 (88)
21,615 (90)
7.15
<.0001
Dyslipidemia
272,651 (84)
19,776 (82)
5.40
<.0001
Peripheral vascular disease
56,113 (17.3)
5355 (22.3)
12.42
<.0001
Renal failure (GFR< 30 or dialysis)
16,379 (5.06)
1761 (7.32)
9.40
<.0001
15.30
<.0001
Glomerular filtration rate
66 (54, 81)
61 (49, 74)
<.0001
Chronic lung disease Severe Moderate
13,365 (4.1) 20,380 (6.3)
1644 (6.8) 2179 (9)
11.92 10.40
Cerebrovascular disease
57,406 (18)
5735 (24)
15.11
<.0001
Transient ischemic attack
15,283 (4.72)
1620 (6.73)
8.68
<.0001
Previous carotid surgery
17,425 (5.38)
1739 (7.23)
7.61
<.0001
Carotid stenosis >75%
10,742 (3.32)
964 (4.01)
3.68
<.0001
8167 (2.5)
993 (4.1)
8.97
<.0001
652 (0.20) 3291 (1.02) 68,921 (21.3)
47 (0.20) 212 (0.88) 7476 (31.1)
Immunosuppressive treatment
<.0001
Preoperative myocardial infarction 6 h >6 h but<24 h 1-21 d
0.13 1.39 22.42
Coronary artery disease Left main 50% 3 vessels 2 vessels 1 vessel
108,365 (33) 249,559 (77) 63,154 (20) 11,205 (3.5)
8713 (36) 18,524 (77) 4720 (20) 815 (3.4)
5.81 0.12 0.31 0.39
<.0001 .77
Left ventricular ejection fraction 55%
182,144 (58)
10,761 (46)
24.45
<.0001
NYHA class IV
9567 (21)
1796 (25)
8.63
<.0001
Valve insufficiency, severe Tricuspid Aortic Mitral
595 (0.18) 216 (0.07) 1050 (0.32)
141 (0.59) 25 (0.10) 189 (0.79)
6.50 1.28 6.21
Aortic stenosis (moderate or severe)
9627 (2.97)
1124 (4.67)
8.87
<.0001
18.16
<.0001
Urgent procedure Surgery years 2013 2012 2011
<.0001
163,417 (50)
14,303 (59)
<.0001 32,815 (10.13) 33,588 (10.37) 36,708 (11.33)
2332 (9.69) 2376 (9.88) 2864 (11.90)
1.47 1.64 1.78 (Continued)
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
3
!ACQ
Body mass index, kg/m
99,345 (31)
AF (N ¼ 24,059)
Acquired
Malaisrie et al
TABLE 1. Continued Variable* 2010 2009 2008 2007 2006 2005
No AF (N ¼ 323,918) 40,521 (12.51) 42,879 (13.24) 44,666 (13.79) 45,181 (13.95) 46,382 (14.46) 728 (0.22)
AF (N ¼ 24,059)
Standardized difference
3336 (13.87) 3399 (14.13) 3351 (13.93) 3259 (13.55) 3071 (12.76) 71 (0.30)
4.01 2.59 0.40 1.17 4.94 1.38
P value
AF, Atrial fibrillation; GFR, glomerular filtration rate; NYHA, New York Heart Association. *Summaries reported for continuous variables are median (first quartile, third quartile); for variables with discrete distributions, we report counts and (%).
!ACQ
continuous variables and most common category for categorical ones. The adjusted survival estimates were computed using all possible combinations of covariates in our dataset, estimating survival curves for each patient in the data set and then averaging over patients in each group of interest to obtain the adjusted curves (with point-wise 95% confidence intervals). Lastly, a possible interaction between AF groups and preoperative Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, Vascular disease, Age 65-74, Sex category (CHA2DS2-VASc) scores16 was evaluated including a multiplicative interaction term in the model and computing hazard ratios for the AF group in each risk group. A possible effect modification by CHA2DS2-VASc scores, categorized in 3 groups: low risk (1-3), moderate risk (4-7) and high risk (7-9) was also evaluated. For the SSE time-to-event analysis, death was considered as a competing risk. Follow-up was censored at death date, end of fee-forservice date or end of study period, whichever came first. Date for inhospital strokes after a procedure was not available, and surgery date was assigned as the event date (23% of all strokes). Cumulative incidence functions for stroke, overall and by AF groups were computed and comparisons among groups were made using Gray tests. For the regression analysis, the Fine and Gray method was used to calculate the subdistribution hazard ratio. The proportional hazards assumption (accounting for the competing risk of death) was tested by plotting Schoenfeld residuals17,18 for each AF group versus log-time and observed no violation. A notsignificant interaction term between AF groups and log-time in the regression model was consistent with this finding. Unadjusted and adjusted subdistribution hazard ratios were then computed for AF versus No AF. The interaction with CHA2DS2-VASc scores groups was evaluated with a multiplicative interaction term. A P value less than 0.05 was considered statistically significant for all tests. All tests were 2-sided. Analyses were performed using SAS (version 9.4, SAS Institute, Cary, NC).
RESULTS Baseline Characteristics Characteristics for patients are summarized in Table 1. Compared to the patients with No AF, patients in the AF group were older and had prior hypertension, renal failure, chronic lung disease, lower ejection fraction and New York Heart Association class III/IV heart failure, but a lower incidence of dyslipidemia. Perioperative Outcomes In-hospital outcomes for the 2 groups are shown in Table 2. Patients in the AF group were more likely to have reoperation, permanent stroke, prolonged ventilation, new renal failure, deep sternal wound infection, and inhospital mortality. Results of regression analysis are shown in Table 3. Compared to those with No AF, the AF group 4
was associated with higher (4.00% vs 1.81%) adjusted 30-day mortality (odds ratio [OR], 1.5; 95% confidence interval [CI], 1.39-1.62; P <.0001) and combined major inhospital morbidity including stroke, renal failure, prolonged ventilation, reoperation for any cause and deep sternal wound infection (OR, 1.32; 95% CI, 1.27-1.37; P < .0001). The AF group had higher adjusted major morbidity in every category except deep sternal wound infection (OR, 1.15; 95% CI, 0.94-1.32; P ¼ .17). Long-Term Survival Unadjusted survival at 30 days, 1 year, 3 years, and 5 years was 98% versus 96%, 94% versus 87%, 88% versus 75%, and 80% versus 63% in patients with No AF versus AF (hazard ratio [HR], 2.09; 95% CI, 2.042.14; P < .0001). Adjusted survival at 30-days, 1-year, 3-years, and 5 years was 98% versus 97%, 94% versus 91%, 88% versus 83%, and 80% versus 73% in patients with No AF versus AF (HR 1.45, 95% CI, 1.41-1.48; P < .0001; Figure 1). A subgroup analysis for mortality by CHA2DS2-VASc scores is displayed in Figure 2. When stratified by scores, 53% of patients had score 1 to 3, 44% of patients had score 4 to 6, and 3% of the overall study population of patients had score 7 to 9. In unadjusted analysis, patients with AF had consistently worse long-term TABLE 2. Perioperative morbidity and mortality (unadjusted in-hospital outcomes by study group) Variable*
frequency
No AF AF (N ¼ 323,918) (N ¼ 24,059) P value
In-hospital mortality
4730 (1.46)
799 (3.32)
<.0001
Operative mortality
5862 (1.81)
962 (4.00)
<.0001
In-hospital complications Reoperation 9259 (2.86) 936 (3.89) Neurologic-permanent 4358 (1.35) 474 (1.97) stroke Prolonged ventilation 25,798 (7.96) 3326 (13.82) New renal failure 10,390 (3.28) 1438 (6.19) Deep sternal wound 966 (0.30) 103 (0.43) infection Major morbidity composite 40,092 (12.38) 4781 (19.87)
<.0001 <.0001 <.0001 <.0001 .0004 <.0001
AF, Atrial fibrillation. *Summaries reported for continuous variables are median (first quartile, third quartile); for variables with discrete distributions, values are n and (%).
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al
Acquired
TABLE 3. Summary of unadjusted and adjusted logistic regression models for in-hospital outcomes of AF group (No AF as reference) Odds ratio
Unadjusted models 95% Confidence interval
P-value
Odds ratio
Adjusted models 95% Confidence interval
P value
In-hospital mortality
2.32
2.15-2.50
<.0001
1.50
1.39-1.62
<.0001
Operative mortality
2.26
2.11-2.43
<.0001
1.47
1.37-1.59
<.0001
Major morbidity
1.76
1.69-1.82
<.0001
1.32
1.27-1.37
<.0001
Prolonged ventilation
1.85
1.77-1.94
<.0001
1.36
1.30-1.42
<.0001
Reoperation: bleeding, graft, valve, other
1.38
1.28-1.48
<.0001
1.18
1.09-1.26
<.0001
Deep sternal wound infection
1.44
1.18-1.76
.0004
1.15
0.94-1.42
.175
Permanent stroke
1.47
1.34-1.62
<.0001
1.20
1.09-1.32
.0003
Renal failure
1.95
1.84-2.06
<.0001
1.35
1.27-1.44
<.0001
survival compared with No AF patients, within each CHA2DS2-VASc score category. At 5 years, the survival probability in the AF group by CHA2DS2-VASc scores was 74.8% (score 1-3), 56.5% (score 4-6), and 41.2% (score 7-9); in comparison, survival in patients with No AF was 86.3%, 73.2%, and 57.2% respectively (log-rank test P value <.001 in all 3 CHA2DS2-VASc score groups). Long-Term Risk of Stroke and Systemic Embolism Unadjusted risk of SSE at 30-days, 1-year, 3-years, and 5 years was 1.7% versus 2.4%, 3.0 versus 4.4%, 5.3% versus 7.8%, and 7.5% versus 10.7%, respectively, in patients with No AF versus AF (HR, 1.43; 95% CI, 1.361.49; P < .0001; Figure 3). Adjusted HR for SSE was 1.24 (95% CI, 1.19-1.30; P < .001). When stratified by CHA2DS2-VASc scores, the 5-year risk of SSE in patients with preoperative AF was 7.9% (score 1-3), 12.2% (score 4-6), and 15.4% (score 7-9; P<.001; Figure 4). In comparison, the 5-year risk of SSE in patients with No AF by
CHA2DS2-VASc scores also increased and was 5.3% (score 1-3), 9.7% (score 4-6), and 16.9% (score 7-9; P <.001). Comment This study confirms prior studies and adds to important late observations. Preoperative AF is common in patients undergoing CABG (10%) and is independently associated with increased in-hospital mortality (OR, 1.5) and major morbidity (OR, 1.32). In this study, patients with preoperative AF were older and had more comorbidities than those without AF, and they had worse long-term mortality (HR, 1.45), even after adjusting for these risk factors. The longterm risk of SSE was significantly worse in patients with preoperative AF (HR, 1.43). Unsurprisingly, they also had higher CHA2DS2-VASc scores (Video 1). Previous reports of patients with preoperative AF undergoing CABG have produced conflicting results for the perioperative period. While most studies have shown that preoperative AF is associated with higher perioperative
FIGURE 1. Survival curves in patients with and without preoperative atrial fibrillation (AF) undergoing coronary artery bypass grafting, Cox regression, unadjusted (hazard ratio, 2.09; 95% confidence interval (CI); P <.0001) and adjusted (hazard ratio, 1.45; 95% CI, 1.41-1.48; P <.0001).
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
5
!ACQ
In-hospital outcomes
Acquired
Malaisrie et al
FIGURE 2. Kaplan-Meier survival curves in patients with preoperative atrial fibrillation (AF) stratified by CHA2DS2-VASc score. CHA2DS2-VASc, Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, VAScular disease, Age 65-74, Sex category.
!ACQ
mortality2,5-8,19 (OR 1.39-2.77), some have shown no difference.3,4 This study documents that preoperative AF is independently associated not only with worse perioperative mortality, but also significant major morbidity (excluding deep sternal wound infection). Prior single-center studies may have failed to detect this difference because of low event rates and inadequate power. Increased perioperative morbidity for stroke, renal failure, and respiratory failure was seen in the AF patients, likely because of the higher thromboembolic events and postoperative low-cardiac output syndrome, perhaps associated with the loss of atrial transport function in the AF patients. Perioperative anticoagulation in the AF patients may explain the higher incidence of reoperation for bleeding. Long-term survival has been shown in many previous reports to be adversely affected in patients with preoperative
FIGURE 3. Cumulative incidence curves of stroke or systemic embolic incidence (in-hospital and stroke readmissions) in patients with and without atrial fibrillation (AF) undergoing coronary artery bypass grafting (P <.0001). Accounting for death as competing risk.
6
AF and several have shown AF to be an independent risk factor with HRs ranging from 1.39 to 2.77.3,4,6,8,19 Similarly, our study showed that preoperative AF increased long-term mortality by almost 50%. Patients with preoperative AF may continue to be at risk for further SSE and thromboembolic events while also remaining at risk for bleeding from anticoagulation therapy. Several additional mechanisms may contribute to the link between AF and long-term mortality, including low cardiac output, tachycardia-mediated cardiomyopathy, and worsening heart failure with loss of atrioventricular synchrony. The observation that late SSE risk is increased in patients with preoperative AF has not been reported previously. This observation supports the hypothesis that AF-related SSE contributes to poor survival. Our study shows that postCABG patients who had AF preoperatively have a continuing increased risk of long-term SSE and mortality risk despite controlling for other risk factors. The analysis of CHA2DS2-VASc is also new to this study. Moreover, the CHA2DS2-VASc score further stratified the risk of SSE concordant with established models, even in patients without preoperative AF.16 A direct link between SSE and mortality has not been previously demonstrated, and further study is justified.
Limitations The increase in relative risk for perioperative morbidity was statistically significant because of the large sample size in this analysis; however, the absolute increase in risk was close to only 1% for 30-day stroke, reoperation, and renal failure. Individually, each complication is of low clinical significance, but the increased absolute risk of combined morbidities was shown to have a greater magnitude of effect size and remains useful for clinical decision
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al
Acquired
making. Likewise, although the absolute increase in risk of 30-day mortality was only about 1%, the absolute increase in risk of death was magnified to 17% at 5 years. Recent clinical practice guidelines20 and expert consensus guidelines21 recommend surgical ablation for AF in patients undergoing cardiac surgery. Further analysis of the comparative effectiveness of concomitant AF ablation may have important implications for this high-risk cohort. This study includes only Medicare-linked patients; therefore, it may be more pertinent to older patients. Because of incomplete linking between the STS and CMS databases, 207,731 patients (37%) were excluded. Limitations to
both registries include lack of data for type of preoperative AF, causality of AF, AF status post procedure, and anticoagulation status. Finally, some patients with AF received surgical AF ablation during CABG and were subsequently excluded from the analysis. CONCLUSIONS Preoperative AF is common in patients undergoing CABG and associated with higher perioperative mortality and morbidity. Long-term survival and risk of SSE were worse even after adjusting for comorbidities. Higher CHA2DS2-VASc scores were associated with worse longterm outcomes including death and SSE. Conflict of Interest Statement Authors have nothing to disclose with regard to commercial support. The authors thank Mr. and Mrs. Timothy Thoelecke for their financial support of the Bluhm Cardiovascular Institute, which made this project possible.
References
VIDEO 1. Dr S. Chris Malaisrie provides an overview of the study aims and outcomes. Video available at: http://www.jtcvsonline.org.
1. Gammie JS, Haddad M, Milford-Beland S, Welke KF, Ferguson TB Jr, O’Brien SM, et al. Atrial fibrillation correction surgery: lessons from the Society of Thoracic Surgeons National Cardiac Database. Ann Thorac Surg. 2008;85: 909-14. 2. Quader MA, McCarthy PM, Gillinov AM, Alster JM, Cosgrove DM III, Lytle BW, et al. Does preoperative atrial fibrillation reduce survival after coronary artery bypass grafting? Ann Thorac Surg. 2004;77:1514-22; discussion 1522-4. 3. Rogers CA, Angelini GD, Culliford LA, Capoun R, Ascione R. Coronary surgery in patients with preexisting chronic atrial fibrillation: early and midterm clinical outcome. Ann Thorac Surg. 2006;81:1676-82. 4. Ngaage DL, Schaff HV, Mullany CJ, Sundt TM III, Dearani JA, Barnes S, et al. Does preoperative atrial fibrillation influence early and late outcomes of coronary artery bypass grafting? J Thorac Cardiovasc Surg. 2007;133:182-9. 5. Ad N, Barnett SD, Haan CK, O’Brien SM, Milford-Beland S, Speir AM. Does preoperative atrial fibrillation increase the risk for mortality and morbidity after coronary artery bypass grafting? J Thorac Cardiovasc Surg. 2009;137:901-6.
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
7
!ACQ
FIGURE 4. Cumulative incidence curves of stroke or systemic embolic incidence in patients with preoperative atrial fibrillation (AF) stratified by CHA2DS2-VASc scores. Adjusted for death as competing risk. CHA2DS2-VASc, Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, VAScular disease, Age 65-74, Sex category.
Acquired
Malaisrie et al
!ACQ
6. Bramer S, van Straten AH, Soliman Hamad MA, Berreklouw E, Martens EJ, Maessen JG. The impact of preoperative atrial fibrillation on early and late mortality after coronary artery bypass grafting. Eur J Cardiothorac Surg. 2010;38:373-9. 7. Al-Sarraf N, Thalib L, Hughes A, Tolan M, Young V, McGovern E. Effect of preoperative atrial fibrillation on postoperative outcome following cardiac surgery. Cardiol Res Pract. 2012;2012:272384. 8. Saxena A, Kapoor J, Dinh DT, Smith JA, Shardey GC, Newcomb AE. Preoperative atrial fibrillation is an independent predictor of worse early and late outcomes after isolated coronary artery bypass graft surgery. J Cardiol. 2015;65:224-9. 9. Badhwar V, Rankin JS, Ad N, Grau-Sepulveda M, Damiano RJ, Gillinov AM, et al. Surgical ablation of atrial fibrillation in the United States: trends and propensity matched outcomes. Ann Thorac Surg. 2017;104:493-500. 10. Jacobs JP, Edwards FH, Shahian DM, Haan CK, Puskas JD, Morales DL, et al. Successful linking of the Society of Thoracic Surgeons adult cardiac surgery database to Centers for Medicare and Medicaid Services Medicare data. Ann Thorac Surg. 2010;90:1150-6; discussion 1156-7. 11. Mann E, Asper F, Durham S. Death information in the research identifiable Medicare data. Available at: https://www.resdac.org/resconnect/articles/117. Accessed September 8, 2017. 12. D’Agostino RS, Jacobs JP, Badhwar V, Paone G, Rankin JS, Han JM, et al. The Society of Thoracic Surgeons Adult Cardiac Surgery Database: 2017 Update on Outcomes and Quality. Ann Thorac Surg. 2017;103:18-24. 13. Shahian DM, Edwards FH, Ferraris VA, Haan CK, Rich JB, Normand SL, et al. Quality measurement in adult cardiac surgery: part 1–Conceptual framework and measure selection. Ann Thorac Surgery. 2007;83(suppl 4):S3-12. 14. O’Brien SM, Shahian DM, DeLong ER, Normand SL, Edwards FH, Ferraris VA, et al. Quality measurement in adult cardiac surgery: part 2–Statistical considerations in composite measure scoring and provider rating. Ann Thorac Surg. 2007; 83(suppl 4):S13-26. 15. Shahian DM, O’Brien SM, Sheng S, Grover FL, Mayer JE, Jacobs JP, et al. Predictors of long-term survival after coronary artery bypass grafting surgery: results from the Society of Thoracic Surgeons Adult Cardiac Surgery Database (the ASCERT study). Circulation. 2012;125:1491-500. 16. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137:263-72. 17. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496-509. 18. Gray RJ. A Class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16:1141-54. 19. Banach M, Goch A, Misztal M, Rysz J, Zaslonka J, Goch JH, et al. Relation between postoperative mortality and atrial fibrillation before surgical revascularization–3-year follow-up. Thorac Cardiovasc Surg. 2008;56:20-3. 20. Badhwar V, Rankin JS, Damiano RJ Jr, Gillinov AM, Bakaeen FG, Edgerton JR, et al. The Society of Thoracic Surgeons 2017 Clinical Practice Guidelines for the Surgical Treatment of Atrial Fibrillation. Ann Thorac Surg. 2017;103:329-41. 21. Ad N, Damiano RJ Jr, Badhwar V, Calkins H, La Meir M, Nitta T, et al. Expert consensus guidelines: examining surgical ablation for atrial fibrillation. J Thorac Cardiovasc Surg. 2017;153:1330-54.e1331.
Key Words: atrial fibrillation, cardiac surgery
Discussion Nahush A. Mokadam. Our next speaker is Chris Malaisrie from Northwestern, and other colleagues speaking on, ‘‘Burden of Preoperative Atrial Fibrillation In Patients Undergoing Coronary Artery Bypass Grafting: An Analysis of the Medicare-Linked Society of Thoracic Surgeons Database.’’ 8
Chris Malaisrie. Thanks, gentlemen. These are our disclosures here. I’d like to thank all the co-authors on this. Patrick McCarthy, Jane Kruse, Adi Andrei, and Jim Cox, who’s now with us at Northwestern, as well as the DCRI Team who managed the data acquisition and analysis, Maria, Dan, and Matt. I would also like to give a special acknowledgment to the Thoelecke family. This project was funded 100% through philanthropy, and from my understanding, this is the first project through the STS/CMS database that has been funded exclusively from philanthropy. Atrial fibrillation, as you know, increases with increasing age. Over the age of 40, there’s an approximately 25% lifetime risk of developing atrial fibrillation. A previous GAMI paper showed the incidence of AF and CABG patients is approximately 6%. The impact of atrial fibrillation on 30-day outcomes after CABG has been conflicting. Some centers have shown that 30-day outcomes are unchanged with preoperative atrial fibrillation, while others have shown that it is detrimental. Late survival has been demonstrated to be uniformly worse in patients with preoperative atrial fibrillation undergoing CABG. Some papers have been able to show that this is an independent risk factor for mortality, but none of these papers have gone on to look at the risk of stroke and systemic embolization. So therefore, the objective was to determine outcomes in a nationally representative population of patients undergoing isolated CABG with and without atrial fibrillation. Excluded from this analysis are patients who underwent surgical atrial fibrillation with Maze procedures and atrial appendage closures. Primary outcome, as you’d expect, is all-cause mortality. Secondary outcomes that we examined were in-hospital mortality, major morbidity, stroke, and systemic embolization. This is the patient flow diagram. We stared out with more than 600,000 patients in the STS database undergoing CABG from 2006 to 2017. After successful linkage with the CMS database, we had about 500,000 patients, and we excluded patients that went on to undergo surgical AF ablation. The final result is approximately 10% of isolated CABG patients in this data set and preoperative atrial fibrillation. Our statistical methods mimicked the statistical methodology that was done in the ASCERT paper. We used Cox regression and KM curves, and we adjusted using multivariable regression. As you quickly look through our baseline characteristics, I will briefly summarize the data in the AF group which, as expected, has more comorbidities, they’re older, with more peripheral vascular disease, more CVAs and strokes, and lower ejection fraction. You can see all the P values are less than 0.0001. This is what you would see when a data set is so large. Results for perioperative mortality in the unadjusted
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Acquired
data set are shown here. Operative mortality is higher: 4% versus 1% in the No AF group. All the 5 major complications are higher in the unadjusted group. When we adjusted for comorbidities, operative mortality remained higher in the AF group with an odds ratio of about 1.50. Four of the major comorbidities remain high. It makes no sense that deep sternal wound infection would be higher because of atrial fibrillation, and that indeed was not the case after adjusting for comorbidities. The central figure from this paper is the overall survival. This figure shows the unadjusted and adjusted curves. The unadjusted hazard ratio for long-term mortality up to 5 years is 2.09. After adjusting for all of the ASCERT baseline characteristics, the hazard ratio is 1.45. That represents an almost 50% increase in mortality in patients that had preoperative atrial fibrillation. The secondary endpoints of stroke and systemic embolism are shown here. Unadjusted hazard ratio for SSE is 1.43. After adjusting, it’s roughly 25% increase in patients with preoperative atrial fibrillation. We went on to further stratify these outcomes in terms of CHADSVASC score, and these are outcomes in patients with preoperative atrial fibrillation stratified by CHADS-VASC score, low intermediate and high CHADS-VASC score, striking 5-year risk of stroke, and systemic embolization in the high CHADS-VASC group of approximately 15.4%. This stratified very well as well, in terms of mortality and if you consider a patient of average age of about 72 to 75 undergoing isolated CABG with preoperative atrial fibrillation. In addition to a high CHADS-VASC score of 7 to 9, their 5-year survival is only 41%. Limitations are that this is an over 65 age group of patients. We lacked data regarding type of atrial fibrillation preoperatively, so cannot classify by paroxysmal atrial fibrillation, longstanding persistent atrial fibrillation. We also do not know the status of postoperative atrial fibrillation including new onset POAF. There was a significant incomplete linking of data between the STS and the CMS database, and about 37% of patients were unable to be linked and again some patients received atrial fibrillation surgery. These patients were excluded from the analysis. Our conclusions are that preoperative atrial fibrillation is common, and about 10% of patients undergoing isolated CABG are associated with higher perioperative mortality and morbidity, excluding deep sternal wound infection. Mortality and risk of stroke and systemic embolization remains high, even adjusting for all of the comorbidities that I listed before. A higher CHADS-VASC score was associated with worse long-term outcomes, including stroke and death. Clearly the implication here is that further analysis of the comparative effectiveness of concomitant AF ablation have important implications for this very-high-risk cohort. Thank you very much. Dr Mokadam. Discussion will be led by Dr Caine.
William T. Caine. Hi. Thanks Dr Mitchell and _____ [blank tape], and I congratulate Dr Malaisrie for this excellent study. You and your colleagues have done a fantastic job, and I appreciate the advance copy of the manuscript. This study, I think, demonstrates a great use of the wealth of data that exists in the STS database, and especially the Medicare linkage that allows long-term mortality statistics to be derived, which are actually contained in the database itself. And you can tell it’s a powerful study, because I noticed in your data that the patients with preoperative AFib were 3 mm taller than the patients who didn’t have AFib, and that was strongly statistically significant, which makes me think that being tall has more disadvantages than just being uncomfortable on airplanes. Anyway, I was surprised to see that in this day and age that only two thirds of the patients in your study had preoperative atrial fibrillation, had no attempt at any procedure, or any ablation, or anything like that to treat their atrial fibrillation, and I wondered if you noticed in your data if in the later years of the study the patients, obviously in later years, if the trend to performing an ablation had grown. Dr Malaisrie. You have to remember that this data set represents all of the cases done in the United States, as you know, and probably in this room most of us would go ahead and perform surgical ablation, and the 30% is about what has been seen in previous data sets from the STS. The undertreatment of atrial fibrillation has decreased over the years. I, unfortunately, do not have that data to show you at this presentation. Dr Caine. Thank you. I also noticed that your hazard ratio is a little bit lower than some of the other published studies that have looked at this question, and I wonder is that because of the wealth or the breadth of your data and better risk adjustment? Or what would you attribute that to? Dr Malaisrie. The Medicare database is an administrative database, and you have to rely on capturing endpoints such as stroke and systemic embolization through ICD-9 codes, so it’s very possible that the rate of stroke is underreported in this certain data set. Certainly, the hard endpoint of mortality, I think, is a reliable one with stroke and systemic embolization likely underreported. Dr Caine. And just the last question. I think we’re all anxious to see if there’s some way to intervene in this process with ablation, if that will improve outcomes and look forward to the results of the data that you’re looking at now, especially the mortality cost of the stroke and systemic embolization issues that you highlighted in your presentation is really staggering. How have you at Northwestern used this data to change your practice, and how should I as a practicing cardiac surgeon implement your data into my day-to-day cardiac surgery practice?
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
9
!ACQ
Malaisrie et al
Acquired
Dr Malaisrie. I think, Bill, you and I are in centers that advocate surgical ablation if it is present at the time of cardiac surgery. I think this data set, I think should help convince the other 7% of surgeons out there that there is a high cost for perioperative or preoperative atrial fibrillation. In particular, patients with very high CHADS-VASC scores, so if you can identify a group of patients that are at risk for stroke and mortality, certainly we would want to bring their survival curve, risk of stroke, down by any way possible. We know that Maze procedure has been shown to be a very safe procedure without increasing perioperative morbidity and mortality. The downside is, of course, cost. I also look forward to the second part of our analysis when we look at the comparative data between the patients that did, in fact, have surgical ablation and that data set is pending. Dr Caine. Thank you. Congratulations again on an excellent _____.
Malaisrie et al
John D. Mitchell. I have a question for you. Great presentation, very clearly done. Are you familiar with how the STS _____ is calculated? And as a result of the data, should it be adjusted? Dr Malaisrie. I think when you actually do these by hand, you’ll notice that you do click on preoperative atrial fibrillation, and STS score and morbidity do change. Dr Mitchell. _____. Dr Malaisrie. Yeah, I think this data set has been incorporated in the current STS PROM. I think there has been another iteration of the calculator recently, and the question is whether or not it is overestimated what we see in practice. I don’t know the answer to that. Dr Mitchell. Thank you.
!ACQ 10
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Acquired
!ACQ
Malaisrie et al
FIGURE E1. Flow diagram for patients undergoing isolated coronary artery bypass grafting as their first cardiovascular surgery with discharge dates from January 1, 2006, to December 31, 2013. CV, Cardiovascular; CABG, coronary artery bypass graft; IABP, intra-aortic balloon pump; PCI, percutaneous coronary intervention; STS, Society of Thoracic Surgeons; CMS, Centers for Medicare and Medicaid; AF, atrial fibrillation.
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
10.e1
Acquired
000
Burden of preoperative atrial fibrillation in patients undergoing coronary artery bypass grafting S. Chris Malaisrie, MD, Patrick M. McCarthy, MD, Jane Kruse, BSN, Roland Matsouaka, PhD, Adin-Cristian Andrei, PhD, Maria V. Grau-Sepulveda, MD, Daniel J. Friedman, MD, James L. Cox, MD, and J. Matthew Brennan, MD, Chicago, Ill, and Durham, NC AF before CABG is associated with higher early and late perioperative mortality and morbidity. Survival and risk of stroke or systemic embolism are worse after adjusting for comorbidities.
!ACQ The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al
Acquired
Burden of preoperative atrial fibrillation in patients undergoing coronary artery bypass grafting S. Chris Malaisrie, MD,a Patrick M. McCarthy, MD,a Jane Kruse, BSN,a Roland Matsouaka, PhD,b Adin-Cristian Andrei, PhD,a Maria V. Grau-Sepulveda, MD,b Daniel J. Friedman, MD,b James L. Cox, MD,a and J. Matthew Brennan, MDb ABSTRACT Background: This study compares early and late outcomes in patients undergoing coronary artery bypass grafting with and without preoperative atrial fibrillation in a contemporary, nationally representative Medicare cohort.
Results: Preoperative atrial fibrillation was associated with a higher adjusted 30day mortality (odds ratio [OR], 1.5; P <.0001) and combined major morbidity including stroke, renal failure, prolonged ventilation, reoperation, and deep sternal wound infection (OR, 1.32; P<.0001). Patients with preoperative atrial fibrillation experienced a higher adjusted long-term risk of all-cause mortality and cumulative risk of stroke and systemic embolism compared to those without atrial fibrillation. At 5 years, the survival probability in the preoperative atrial fibrillation versus no atrial fibrillation groups stratified by CHA2DS2-VASc scores was 74.8% versus 86.3% (score 1-3), 56.5% versus 73.2% (score 4-6), and 41.2% versus 57.2% (score 7-9; all P <.001).
Survival is worse in patients with preoperative AF compared to those with no AF (P < .001). Central Message AF before CABG is associated with higher early and late perioperative mortality and morbidity. Survival and risk of stroke or systemic embolism are worse after adjusting for comorbidities.
!ACQ
Methods: In the Medicare-Linked Society of Thoracic Surgeons database, 361,138 patients underwent isolated coronary artery bypass from 2006 to 2013, of whom 37,220 (10.3%) had preoperative atrial fibrillation; 13,161 (35.4%) were treated with surgical ablation and were excluded. Generalized estimating equations were used to compare 30-day mortality and morbidity. Long-term survival was summarized using Kaplan-Meier curves and Cox regression models. Stroke and systemic embolism incidence was modeled using the Fine-Gray model and the CHA2DS2-VASc score was used to analyze stroke risk. Median follow-up was 4 years.
Perspective Preoperative atrial fibrillation is an independent risk factor for worse outcomes after coronary artery bypass surgery. Further analysis of the comparative effectiveness of concomitant atrial fibrillation ablation has important implications for this high-risk cohort.
See Editorial Commentary page XXX.
Conclusions: Preoperative atrial fibrillation is independently associated with worse early and late postoperative outcomes. CHA2DS2-VASc stratifies risk, even in those without preoperative atrial fibrillation. (J Thorac Cardiovasc Surg 2018;-:1-10)
The incidence of atrial fibrillation in patients with coronary artery disease (CAD) requiring coronary artery bypass From the aDivision of Cardiac Surgery, Northwestern University, Bluhm Cardiovascular Institute, Chicago, Ill; and bDuke Clinical Research Institute, Duke University School of Medicine, Durham, NC. This work was supported by institutional funding from Northwestern University. Read at the 43rd Annual Meeting of The Western Thoracic Surgical Association, Colorado Springs, Colorado, June 21-24, 2017. Received for publication June 21, 2017; revisions received Dec 5, 2017; accepted for publication Jan 9, 2018. Address for reprints: S. Chris Malaisrie, MD, Division of Cardiac Surgery at Northwestern University Feinberg School of Medicine and Northwestern, Bluhm Cardiovascular Institute, 201 East Huron St, Suite 11-140, Chicago, IL 60611-2908 (E-mail: [email protected]). 0022-5223/$36.00 Copyright Ó 2018 by The American Association for Thoracic Surgery https://doi.org/10.1016/j.jtcvs.2018.01.069
grafting (CABG) is approximately 6% according to a historical Society of Thoracic Surgeons (STS) report.1 Published reports on the association between preoperative atrial fibrillation (AF) and post-CABG outcomes have shown worse perioperative mortality and long-term survival.2-8 Previous STS database studies have shown increased perioperative mortality and stroke in patients with preoperative AF, but long-term outcomes were not studied.9 Scanning this QR code will take you to a supplemental figure and video for the article.
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
1
Acquired
!ACQ
Abbreviations and Acronyms ASCERT ¼ American College of Cardiology Foundation–Society of Thoracic Surgeons Collaboration on the Comparative Effectiveness of Revascularization Strategies AF ¼ atrial fibrillation CABG ¼ coronary artery bypass graft CAD ¼ coronary artery disease CHA2DS2-VASc ¼ Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, VAScular disease, Age 65-74, Sex category CI ¼ confidence interval CMS ¼ Centers for Medicare and Medicaid HR ¼ hazard ratio OR ¼ odds ratio SSE ¼ stroke or systemic embolism STS ¼ Society of Thoracic Surgeons
The STS database recent linkage to the Centers for Medicare and Medicaid Services (CMS) database allows for study of important long-term endpoints beyond perioperative outcomes provided in the STS National Cardiac Database. Using CMS-linked STS data, this study sought to examine early and late outcomes in 2 groups of first cardiac surgery CABG patients: (1) those with preoperative AF but no surgical AF treatment, and (2) those without preoperative AF. PATIENTS AND METHODS Data for all patients were obtained from the STS Adult Cardiac Surgery Database (versions 2.52, 2.61, and 2.73) for patients discharged between January 1, 2006, and December 31, 2013, and who could be linked to CMS data using a validated deterministic matching algorithm.10
Study Population Medicare recipients age 65 years or older and undergoing CABG as a first cardiac surgery who had complete information on preoperative history of atrial arrhythmias were included. Between January 2006 and December 2013, there were 688,466 isolated CABG surgery procedures reported in the STS database in patients age 65 years or older. Of those, 119,597 patients were excluded because of prior cardiac surgery (n ¼ 44,671), emergent/salvage status–cardiogenic shock/resuscitation (n ¼ 25,542), preoperative intra-aortic balloon pump/inotropes (n ¼ 44,977), history of endocarditis (n ¼ 637), and missing or inconsistent data on arrhythmia, number of diseased vessels, gender, or any of the previous fields considered for exclusion (n ¼ 3770). Of the resulting population, additional exclusions were made for 207,553 with unlinked CMS data, 178 had more than 1 admission on the same day, and 13,161 had surgical ablation. The resulting 347,977 patients with linked CMS data (61%) were categorized according to their preoperative AF status: (1) 89.7% (n ¼ 323,918) no preoperative
2
Malaisrie et al
AF (No AF) or (2) 6.7% (n ¼ 24,059) preoperative AF (AF group) and analyzed (Figure E1).
Study Outcomes The primary outcome was all-cause mortality after CABG. This was defined using STS registry data for in-hospital deaths and the linked Medicare Denominator File for postdischarge deaths.11 Secondary outcomes included (1) stroke or systemic embolism (SSE; ischemic stroke, hemorrhagic stroke, transient ischemic attack, or systemic arterial embolism), (2) in-hospital mortality, and (3) a composite of inhospital major morbidity.12 Incident SSE was defined by using STS registry data to account for inhospital strokes and Medicare Part A data to identify subsequent rehospitalizations with SSE as a primary diagnosis (International Classification of Diseases, 9th revision, codes: 433.1, 434.x1, 430, 431, 432.0, 432.1, 432.9, 444,x, 435.x). The STS database does not include the exact date of in-hospital strokes; for the purpose of time-to-event analyses, we assigned the date of in-hospital stroke to be the date of the index operation. In-hospital major morbidity, a previously defined composite,13,14 referred to any of the following postprocedure complications: permanent stroke, new cases of renal failure, prolonged ventilation (ventilation longer than 24 hours after surgery), reoperation for cardiac reasons (graft dysfunction, bleeding, valve dysfunction or other), and deep sternal wound infection.
Statistical Methods The overall study population and the subgroups of interest (No AF vs AF) were summarized using medians and interquartile ranges (25th and 75th percentiles) for continuous variables and frequency counts and percentages for categorical variables. Differences in distributions among patients with No AF versus AF were evaluated with Wilcoxon and Pearson chi-square tests, respectively. Logistic regression models were developed to estimate the association between in-hospital outcomes and AF versus No AF. The generalized estimated equations method with an exchangeable correlation structure was used to compute confidence limits that account for hospital clustering of patients. Adjusted models included predictors of long-term survival after CABG identified with the validated American College of Cardiology Foundation–Society of Thoracic Surgeons Collaboration on the Comparative Effectiveness of Revascularization Strategies (ASCERT) model15: age, ejection fraction, year of surgery, renal failure (glomerular filtration rate <30 or dialysis), glomerular filtration rate, body mass index, height, sex, race, current smoking, chronic heart failure/New York Heart Association Class IV, urgent operative status, number of diseased vessels, diabetes, previous carotid surgery, carotid stenosis > 75%, transient ischemic attack, chronic lung disease moderate/severe, left main disease, unstable angina, peripheral vascular disease, immunosuppressant medication, hypertension, valve insufficiency (tricuspid, aortic, mitral) moderate/severe, aortic stenosis gradient, preoperative myocardial infarction (less than 6 hours before, between 6 and 24 hours, between 1 and 21 days). These variables were also used for all adjusted time-to-event analyses. Time-to-event analysis was used to compare long-term survival and SSE occurrence by group. For survival, patient follow-up was censored at the end of study period (January 1, 2014). Product-limit Kaplan Meier survival estimates were computed for each group and comparisons were made with log-rank tests. Cox regression models were used to compute hazard ratios for the AF versus No AF, both unadjusted and adjusted by other survival predictors. A robust sandwich covariance estimation to compute 95% confidence intervals was used to account for correlation of patients’ failure times within the same hospital. The proportional hazard assumption was tested using log-log survival plots (log-log survival vs log-time) for each group and the results indicated no violation of this assumption. Consistent with the validated STS risk models,15 missing values (<3%) were imputed with relevant groups-specific medians for
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al
Acquired
TABLE 1. Baseline characteristics of patients in No AF or AF groups No AF (N ¼ 323,918)
Variable* Age, y
72 (68, 77)
Female sex 2
28.09 (25, 32)
Height, cm
172.7 (163, 178)
75 (71, 80) 6920 (29) 28.10 (25, 32) 173 (165, 180)
Standardized difference
P value
39.94
<.0001
4.17
<.0001
1.93
.954
12.06
<.0001 <.0001
Race Other Asian Hispanic Black Caucasian Missing Current smoking
6725 (2.08) 5138 (1.59) 11,218 (3.46) 16,326 (5.04) 283,245 (87.44) 1266 (0.39)
428 (1.78) 283 (1.18) 548 (2.28) 751 (3.12) 21,952 (91.24) 97 (0.40)
2.16 3.51 7.10 9.71 12.33 0.20
46,294 (14.29)
3111 (12.93)
4.00
<.0001
Diabetes
129,348 (40)
9903 (41)
2.50
<.0001
Hypertension
283,689 (88)
21,615 (90)
7.15
<.0001
Dyslipidemia
272,651 (84)
19,776 (82)
5.40
<.0001
Peripheral vascular disease
56,113 (17.3)
5355 (22.3)
12.42
<.0001
Renal failure (GFR< 30 or dialysis)
16,379 (5.06)
1761 (7.32)
9.40
<.0001
15.30
<.0001
Glomerular filtration rate
66 (54, 81)
61 (49, 74)
<.0001
Chronic lung disease Severe Moderate
13,365 (4.1) 20,380 (6.3)
1644 (6.8) 2179 (9)
11.92 10.40
Cerebrovascular disease
57,406 (18)
5735 (24)
15.11
<.0001
Transient ischemic attack
15,283 (4.72)
1620 (6.73)
8.68
<.0001
Previous carotid surgery
17,425 (5.38)
1739 (7.23)
7.61
<.0001
Carotid stenosis >75%
10,742 (3.32)
964 (4.01)
3.68
<.0001
8167 (2.5)
993 (4.1)
8.97
<.0001
652 (0.20) 3291 (1.02) 68,921 (21.3)
47 (0.20) 212 (0.88) 7476 (31.1)
Immunosuppressive treatment
<.0001
Preoperative myocardial infarction 6 h >6 h but<24 h 1-21 d
0.13 1.39 22.42
Coronary artery disease Left main 50% 3 vessels 2 vessels 1 vessel
108,365 (33) 249,559 (77) 63,154 (20) 11,205 (3.5)
8713 (36) 18,524 (77) 4720 (20) 815 (3.4)
5.81 0.12 0.31 0.39
<.0001 .77
Left ventricular ejection fraction 55%
182,144 (58)
10,761 (46)
24.45
<.0001
NYHA class IV
9567 (21)
1796 (25)
8.63
<.0001
Valve insufficiency, severe Tricuspid Aortic Mitral
595 (0.18) 216 (0.07) 1050 (0.32)
141 (0.59) 25 (0.10) 189 (0.79)
6.50 1.28 6.21
Aortic stenosis (moderate or severe)
9627 (2.97)
1124 (4.67)
8.87
<.0001
18.16
<.0001
Urgent procedure Surgery years 2013 2012 2011
<.0001
163,417 (50)
14,303 (59)
<.0001 32,815 (10.13) 33,588 (10.37) 36,708 (11.33)
2332 (9.69) 2376 (9.88) 2864 (11.90)
1.47 1.64 1.78 (Continued)
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
3
!ACQ
Body mass index, kg/m
99,345 (31)
AF (N ¼ 24,059)
Acquired
Malaisrie et al
TABLE 1. Continued Variable* 2010 2009 2008 2007 2006 2005
No AF (N ¼ 323,918) 40,521 (12.51) 42,879 (13.24) 44,666 (13.79) 45,181 (13.95) 46,382 (14.46) 728 (0.22)
AF (N ¼ 24,059)
Standardized difference
3336 (13.87) 3399 (14.13) 3351 (13.93) 3259 (13.55) 3071 (12.76) 71 (0.30)
4.01 2.59 0.40 1.17 4.94 1.38
P value
AF, Atrial fibrillation; GFR, glomerular filtration rate; NYHA, New York Heart Association. *Summaries reported for continuous variables are median (first quartile, third quartile); for variables with discrete distributions, we report counts and (%).
!ACQ
continuous variables and most common category for categorical ones. The adjusted survival estimates were computed using all possible combinations of covariates in our dataset, estimating survival curves for each patient in the data set and then averaging over patients in each group of interest to obtain the adjusted curves (with point-wise 95% confidence intervals). Lastly, a possible interaction between AF groups and preoperative Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, Vascular disease, Age 65-74, Sex category (CHA2DS2-VASc) scores16 was evaluated including a multiplicative interaction term in the model and computing hazard ratios for the AF group in each risk group. A possible effect modification by CHA2DS2-VASc scores, categorized in 3 groups: low risk (1-3), moderate risk (4-7) and high risk (7-9) was also evaluated. For the SSE time-to-event analysis, death was considered as a competing risk. Follow-up was censored at death date, end of fee-forservice date or end of study period, whichever came first. Date for inhospital strokes after a procedure was not available, and surgery date was assigned as the event date (23% of all strokes). Cumulative incidence functions for stroke, overall and by AF groups were computed and comparisons among groups were made using Gray tests. For the regression analysis, the Fine and Gray method was used to calculate the subdistribution hazard ratio. The proportional hazards assumption (accounting for the competing risk of death) was tested by plotting Schoenfeld residuals17,18 for each AF group versus log-time and observed no violation. A notsignificant interaction term between AF groups and log-time in the regression model was consistent with this finding. Unadjusted and adjusted subdistribution hazard ratios were then computed for AF versus No AF. The interaction with CHA2DS2-VASc scores groups was evaluated with a multiplicative interaction term. A P value less than 0.05 was considered statistically significant for all tests. All tests were 2-sided. Analyses were performed using SAS (version 9.4, SAS Institute, Cary, NC).
RESULTS Baseline Characteristics Characteristics for patients are summarized in Table 1. Compared to the patients with No AF, patients in the AF group were older and had prior hypertension, renal failure, chronic lung disease, lower ejection fraction and New York Heart Association class III/IV heart failure, but a lower incidence of dyslipidemia. Perioperative Outcomes In-hospital outcomes for the 2 groups are shown in Table 2. Patients in the AF group were more likely to have reoperation, permanent stroke, prolonged ventilation, new renal failure, deep sternal wound infection, and inhospital mortality. Results of regression analysis are shown in Table 3. Compared to those with No AF, the AF group 4
was associated with higher (4.00% vs 1.81%) adjusted 30-day mortality (odds ratio [OR], 1.5; 95% confidence interval [CI], 1.39-1.62; P <.0001) and combined major inhospital morbidity including stroke, renal failure, prolonged ventilation, reoperation for any cause and deep sternal wound infection (OR, 1.32; 95% CI, 1.27-1.37; P < .0001). The AF group had higher adjusted major morbidity in every category except deep sternal wound infection (OR, 1.15; 95% CI, 0.94-1.32; P ¼ .17). Long-Term Survival Unadjusted survival at 30 days, 1 year, 3 years, and 5 years was 98% versus 96%, 94% versus 87%, 88% versus 75%, and 80% versus 63% in patients with No AF versus AF (hazard ratio [HR], 2.09; 95% CI, 2.042.14; P < .0001). Adjusted survival at 30-days, 1-year, 3-years, and 5 years was 98% versus 97%, 94% versus 91%, 88% versus 83%, and 80% versus 73% in patients with No AF versus AF (HR 1.45, 95% CI, 1.41-1.48; P < .0001; Figure 1). A subgroup analysis for mortality by CHA2DS2-VASc scores is displayed in Figure 2. When stratified by scores, 53% of patients had score 1 to 3, 44% of patients had score 4 to 6, and 3% of the overall study population of patients had score 7 to 9. In unadjusted analysis, patients with AF had consistently worse long-term TABLE 2. Perioperative morbidity and mortality (unadjusted in-hospital outcomes by study group) Variable*
frequency
No AF AF (N ¼ 323,918) (N ¼ 24,059) P value
In-hospital mortality
4730 (1.46)
799 (3.32)
<.0001
Operative mortality
5862 (1.81)
962 (4.00)
<.0001
In-hospital complications Reoperation 9259 (2.86) 936 (3.89) Neurologic-permanent 4358 (1.35) 474 (1.97) stroke Prolonged ventilation 25,798 (7.96) 3326 (13.82) New renal failure 10,390 (3.28) 1438 (6.19) Deep sternal wound 966 (0.30) 103 (0.43) infection Major morbidity composite 40,092 (12.38) 4781 (19.87)
<.0001 <.0001 <.0001 <.0001 .0004 <.0001
AF, Atrial fibrillation. *Summaries reported for continuous variables are median (first quartile, third quartile); for variables with discrete distributions, values are n and (%).
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al
Acquired
TABLE 3. Summary of unadjusted and adjusted logistic regression models for in-hospital outcomes of AF group (No AF as reference) Odds ratio
Unadjusted models 95% Confidence interval
P-value
Odds ratio
Adjusted models 95% Confidence interval
P value
In-hospital mortality
2.32
2.15-2.50
<.0001
1.50
1.39-1.62
<.0001
Operative mortality
2.26
2.11-2.43
<.0001
1.47
1.37-1.59
<.0001
Major morbidity
1.76
1.69-1.82
<.0001
1.32
1.27-1.37
<.0001
Prolonged ventilation
1.85
1.77-1.94
<.0001
1.36
1.30-1.42
<.0001
Reoperation: bleeding, graft, valve, other
1.38
1.28-1.48
<.0001
1.18
1.09-1.26
<.0001
Deep sternal wound infection
1.44
1.18-1.76
.0004
1.15
0.94-1.42
.175
Permanent stroke
1.47
1.34-1.62
<.0001
1.20
1.09-1.32
.0003
Renal failure
1.95
1.84-2.06
<.0001
1.35
1.27-1.44
<.0001
survival compared with No AF patients, within each CHA2DS2-VASc score category. At 5 years, the survival probability in the AF group by CHA2DS2-VASc scores was 74.8% (score 1-3), 56.5% (score 4-6), and 41.2% (score 7-9); in comparison, survival in patients with No AF was 86.3%, 73.2%, and 57.2% respectively (log-rank test P value <.001 in all 3 CHA2DS2-VASc score groups). Long-Term Risk of Stroke and Systemic Embolism Unadjusted risk of SSE at 30-days, 1-year, 3-years, and 5 years was 1.7% versus 2.4%, 3.0 versus 4.4%, 5.3% versus 7.8%, and 7.5% versus 10.7%, respectively, in patients with No AF versus AF (HR, 1.43; 95% CI, 1.361.49; P < .0001; Figure 3). Adjusted HR for SSE was 1.24 (95% CI, 1.19-1.30; P < .001). When stratified by CHA2DS2-VASc scores, the 5-year risk of SSE in patients with preoperative AF was 7.9% (score 1-3), 12.2% (score 4-6), and 15.4% (score 7-9; P<.001; Figure 4). In comparison, the 5-year risk of SSE in patients with No AF by
CHA2DS2-VASc scores also increased and was 5.3% (score 1-3), 9.7% (score 4-6), and 16.9% (score 7-9; P <.001). Comment This study confirms prior studies and adds to important late observations. Preoperative AF is common in patients undergoing CABG (10%) and is independently associated with increased in-hospital mortality (OR, 1.5) and major morbidity (OR, 1.32). In this study, patients with preoperative AF were older and had more comorbidities than those without AF, and they had worse long-term mortality (HR, 1.45), even after adjusting for these risk factors. The longterm risk of SSE was significantly worse in patients with preoperative AF (HR, 1.43). Unsurprisingly, they also had higher CHA2DS2-VASc scores (Video 1). Previous reports of patients with preoperative AF undergoing CABG have produced conflicting results for the perioperative period. While most studies have shown that preoperative AF is associated with higher perioperative
FIGURE 1. Survival curves in patients with and without preoperative atrial fibrillation (AF) undergoing coronary artery bypass grafting, Cox regression, unadjusted (hazard ratio, 2.09; 95% confidence interval (CI); P <.0001) and adjusted (hazard ratio, 1.45; 95% CI, 1.41-1.48; P <.0001).
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
5
!ACQ
In-hospital outcomes
Acquired
Malaisrie et al
FIGURE 2. Kaplan-Meier survival curves in patients with preoperative atrial fibrillation (AF) stratified by CHA2DS2-VASc score. CHA2DS2-VASc, Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, VAScular disease, Age 65-74, Sex category.
!ACQ
mortality2,5-8,19 (OR 1.39-2.77), some have shown no difference.3,4 This study documents that preoperative AF is independently associated not only with worse perioperative mortality, but also significant major morbidity (excluding deep sternal wound infection). Prior single-center studies may have failed to detect this difference because of low event rates and inadequate power. Increased perioperative morbidity for stroke, renal failure, and respiratory failure was seen in the AF patients, likely because of the higher thromboembolic events and postoperative low-cardiac output syndrome, perhaps associated with the loss of atrial transport function in the AF patients. Perioperative anticoagulation in the AF patients may explain the higher incidence of reoperation for bleeding. Long-term survival has been shown in many previous reports to be adversely affected in patients with preoperative
FIGURE 3. Cumulative incidence curves of stroke or systemic embolic incidence (in-hospital and stroke readmissions) in patients with and without atrial fibrillation (AF) undergoing coronary artery bypass grafting (P <.0001). Accounting for death as competing risk.
6
AF and several have shown AF to be an independent risk factor with HRs ranging from 1.39 to 2.77.3,4,6,8,19 Similarly, our study showed that preoperative AF increased long-term mortality by almost 50%. Patients with preoperative AF may continue to be at risk for further SSE and thromboembolic events while also remaining at risk for bleeding from anticoagulation therapy. Several additional mechanisms may contribute to the link between AF and long-term mortality, including low cardiac output, tachycardia-mediated cardiomyopathy, and worsening heart failure with loss of atrioventricular synchrony. The observation that late SSE risk is increased in patients with preoperative AF has not been reported previously. This observation supports the hypothesis that AF-related SSE contributes to poor survival. Our study shows that postCABG patients who had AF preoperatively have a continuing increased risk of long-term SSE and mortality risk despite controlling for other risk factors. The analysis of CHA2DS2-VASc is also new to this study. Moreover, the CHA2DS2-VASc score further stratified the risk of SSE concordant with established models, even in patients without preoperative AF.16 A direct link between SSE and mortality has not been previously demonstrated, and further study is justified.
Limitations The increase in relative risk for perioperative morbidity was statistically significant because of the large sample size in this analysis; however, the absolute increase in risk was close to only 1% for 30-day stroke, reoperation, and renal failure. Individually, each complication is of low clinical significance, but the increased absolute risk of combined morbidities was shown to have a greater magnitude of effect size and remains useful for clinical decision
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al
Acquired
making. Likewise, although the absolute increase in risk of 30-day mortality was only about 1%, the absolute increase in risk of death was magnified to 17% at 5 years. Recent clinical practice guidelines20 and expert consensus guidelines21 recommend surgical ablation for AF in patients undergoing cardiac surgery. Further analysis of the comparative effectiveness of concomitant AF ablation may have important implications for this high-risk cohort. This study includes only Medicare-linked patients; therefore, it may be more pertinent to older patients. Because of incomplete linking between the STS and CMS databases, 207,731 patients (37%) were excluded. Limitations to
both registries include lack of data for type of preoperative AF, causality of AF, AF status post procedure, and anticoagulation status. Finally, some patients with AF received surgical AF ablation during CABG and were subsequently excluded from the analysis. CONCLUSIONS Preoperative AF is common in patients undergoing CABG and associated with higher perioperative mortality and morbidity. Long-term survival and risk of SSE were worse even after adjusting for comorbidities. Higher CHA2DS2-VASc scores were associated with worse longterm outcomes including death and SSE. Conflict of Interest Statement Authors have nothing to disclose with regard to commercial support. The authors thank Mr. and Mrs. Timothy Thoelecke for their financial support of the Bluhm Cardiovascular Institute, which made this project possible.
References
VIDEO 1. Dr S. Chris Malaisrie provides an overview of the study aims and outcomes. Video available at: http://www.jtcvsonline.org.
1. Gammie JS, Haddad M, Milford-Beland S, Welke KF, Ferguson TB Jr, O’Brien SM, et al. Atrial fibrillation correction surgery: lessons from the Society of Thoracic Surgeons National Cardiac Database. Ann Thorac Surg. 2008;85: 909-14. 2. Quader MA, McCarthy PM, Gillinov AM, Alster JM, Cosgrove DM III, Lytle BW, et al. Does preoperative atrial fibrillation reduce survival after coronary artery bypass grafting? Ann Thorac Surg. 2004;77:1514-22; discussion 1522-4. 3. Rogers CA, Angelini GD, Culliford LA, Capoun R, Ascione R. Coronary surgery in patients with preexisting chronic atrial fibrillation: early and midterm clinical outcome. Ann Thorac Surg. 2006;81:1676-82. 4. Ngaage DL, Schaff HV, Mullany CJ, Sundt TM III, Dearani JA, Barnes S, et al. Does preoperative atrial fibrillation influence early and late outcomes of coronary artery bypass grafting? J Thorac Cardiovasc Surg. 2007;133:182-9. 5. Ad N, Barnett SD, Haan CK, O’Brien SM, Milford-Beland S, Speir AM. Does preoperative atrial fibrillation increase the risk for mortality and morbidity after coronary artery bypass grafting? J Thorac Cardiovasc Surg. 2009;137:901-6.
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
7
!ACQ
FIGURE 4. Cumulative incidence curves of stroke or systemic embolic incidence in patients with preoperative atrial fibrillation (AF) stratified by CHA2DS2-VASc scores. Adjusted for death as competing risk. CHA2DS2-VASc, Congestive heart failure, Hypertension, Age 75, Diabetes, Stroke, VAScular disease, Age 65-74, Sex category.
Acquired
Malaisrie et al
!ACQ
6. Bramer S, van Straten AH, Soliman Hamad MA, Berreklouw E, Martens EJ, Maessen JG. The impact of preoperative atrial fibrillation on early and late mortality after coronary artery bypass grafting. Eur J Cardiothorac Surg. 2010;38:373-9. 7. Al-Sarraf N, Thalib L, Hughes A, Tolan M, Young V, McGovern E. Effect of preoperative atrial fibrillation on postoperative outcome following cardiac surgery. Cardiol Res Pract. 2012;2012:272384. 8. Saxena A, Kapoor J, Dinh DT, Smith JA, Shardey GC, Newcomb AE. Preoperative atrial fibrillation is an independent predictor of worse early and late outcomes after isolated coronary artery bypass graft surgery. J Cardiol. 2015;65:224-9. 9. Badhwar V, Rankin JS, Ad N, Grau-Sepulveda M, Damiano RJ, Gillinov AM, et al. Surgical ablation of atrial fibrillation in the United States: trends and propensity matched outcomes. Ann Thorac Surg. 2017;104:493-500. 10. Jacobs JP, Edwards FH, Shahian DM, Haan CK, Puskas JD, Morales DL, et al. Successful linking of the Society of Thoracic Surgeons adult cardiac surgery database to Centers for Medicare and Medicaid Services Medicare data. Ann Thorac Surg. 2010;90:1150-6; discussion 1156-7. 11. Mann E, Asper F, Durham S. Death information in the research identifiable Medicare data. Available at: https://www.resdac.org/resconnect/articles/117. Accessed September 8, 2017. 12. D’Agostino RS, Jacobs JP, Badhwar V, Paone G, Rankin JS, Han JM, et al. The Society of Thoracic Surgeons Adult Cardiac Surgery Database: 2017 Update on Outcomes and Quality. Ann Thorac Surg. 2017;103:18-24. 13. Shahian DM, Edwards FH, Ferraris VA, Haan CK, Rich JB, Normand SL, et al. Quality measurement in adult cardiac surgery: part 1–Conceptual framework and measure selection. Ann Thorac Surgery. 2007;83(suppl 4):S3-12. 14. O’Brien SM, Shahian DM, DeLong ER, Normand SL, Edwards FH, Ferraris VA, et al. Quality measurement in adult cardiac surgery: part 2–Statistical considerations in composite measure scoring and provider rating. Ann Thorac Surg. 2007; 83(suppl 4):S13-26. 15. Shahian DM, O’Brien SM, Sheng S, Grover FL, Mayer JE, Jacobs JP, et al. Predictors of long-term survival after coronary artery bypass grafting surgery: results from the Society of Thoracic Surgeons Adult Cardiac Surgery Database (the ASCERT study). Circulation. 2012;125:1491-500. 16. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137:263-72. 17. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496-509. 18. Gray RJ. A Class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988;16:1141-54. 19. Banach M, Goch A, Misztal M, Rysz J, Zaslonka J, Goch JH, et al. Relation between postoperative mortality and atrial fibrillation before surgical revascularization–3-year follow-up. Thorac Cardiovasc Surg. 2008;56:20-3. 20. Badhwar V, Rankin JS, Damiano RJ Jr, Gillinov AM, Bakaeen FG, Edgerton JR, et al. The Society of Thoracic Surgeons 2017 Clinical Practice Guidelines for the Surgical Treatment of Atrial Fibrillation. Ann Thorac Surg. 2017;103:329-41. 21. Ad N, Damiano RJ Jr, Badhwar V, Calkins H, La Meir M, Nitta T, et al. Expert consensus guidelines: examining surgical ablation for atrial fibrillation. J Thorac Cardiovasc Surg. 2017;153:1330-54.e1331.
Key Words: atrial fibrillation, cardiac surgery
Discussion Nahush A. Mokadam. Our next speaker is Chris Malaisrie from Northwestern, and other colleagues speaking on, ‘‘Burden of Preoperative Atrial Fibrillation In Patients Undergoing Coronary Artery Bypass Grafting: An Analysis of the Medicare-Linked Society of Thoracic Surgeons Database.’’ 8
Chris Malaisrie. Thanks, gentlemen. These are our disclosures here. I’d like to thank all the co-authors on this. Patrick McCarthy, Jane Kruse, Adi Andrei, and Jim Cox, who’s now with us at Northwestern, as well as the DCRI Team who managed the data acquisition and analysis, Maria, Dan, and Matt. I would also like to give a special acknowledgment to the Thoelecke family. This project was funded 100% through philanthropy, and from my understanding, this is the first project through the STS/CMS database that has been funded exclusively from philanthropy. Atrial fibrillation, as you know, increases with increasing age. Over the age of 40, there’s an approximately 25% lifetime risk of developing atrial fibrillation. A previous GAMI paper showed the incidence of AF and CABG patients is approximately 6%. The impact of atrial fibrillation on 30-day outcomes after CABG has been conflicting. Some centers have shown that 30-day outcomes are unchanged with preoperative atrial fibrillation, while others have shown that it is detrimental. Late survival has been demonstrated to be uniformly worse in patients with preoperative atrial fibrillation undergoing CABG. Some papers have been able to show that this is an independent risk factor for mortality, but none of these papers have gone on to look at the risk of stroke and systemic embolization. So therefore, the objective was to determine outcomes in a nationally representative population of patients undergoing isolated CABG with and without atrial fibrillation. Excluded from this analysis are patients who underwent surgical atrial fibrillation with Maze procedures and atrial appendage closures. Primary outcome, as you’d expect, is all-cause mortality. Secondary outcomes that we examined were in-hospital mortality, major morbidity, stroke, and systemic embolization. This is the patient flow diagram. We stared out with more than 600,000 patients in the STS database undergoing CABG from 2006 to 2017. After successful linkage with the CMS database, we had about 500,000 patients, and we excluded patients that went on to undergo surgical AF ablation. The final result is approximately 10% of isolated CABG patients in this data set and preoperative atrial fibrillation. Our statistical methods mimicked the statistical methodology that was done in the ASCERT paper. We used Cox regression and KM curves, and we adjusted using multivariable regression. As you quickly look through our baseline characteristics, I will briefly summarize the data in the AF group which, as expected, has more comorbidities, they’re older, with more peripheral vascular disease, more CVAs and strokes, and lower ejection fraction. You can see all the P values are less than 0.0001. This is what you would see when a data set is so large. Results for perioperative mortality in the unadjusted
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Acquired
data set are shown here. Operative mortality is higher: 4% versus 1% in the No AF group. All the 5 major complications are higher in the unadjusted group. When we adjusted for comorbidities, operative mortality remained higher in the AF group with an odds ratio of about 1.50. Four of the major comorbidities remain high. It makes no sense that deep sternal wound infection would be higher because of atrial fibrillation, and that indeed was not the case after adjusting for comorbidities. The central figure from this paper is the overall survival. This figure shows the unadjusted and adjusted curves. The unadjusted hazard ratio for long-term mortality up to 5 years is 2.09. After adjusting for all of the ASCERT baseline characteristics, the hazard ratio is 1.45. That represents an almost 50% increase in mortality in patients that had preoperative atrial fibrillation. The secondary endpoints of stroke and systemic embolism are shown here. Unadjusted hazard ratio for SSE is 1.43. After adjusting, it’s roughly 25% increase in patients with preoperative atrial fibrillation. We went on to further stratify these outcomes in terms of CHADSVASC score, and these are outcomes in patients with preoperative atrial fibrillation stratified by CHADS-VASC score, low intermediate and high CHADS-VASC score, striking 5-year risk of stroke, and systemic embolization in the high CHADS-VASC group of approximately 15.4%. This stratified very well as well, in terms of mortality and if you consider a patient of average age of about 72 to 75 undergoing isolated CABG with preoperative atrial fibrillation. In addition to a high CHADS-VASC score of 7 to 9, their 5-year survival is only 41%. Limitations are that this is an over 65 age group of patients. We lacked data regarding type of atrial fibrillation preoperatively, so cannot classify by paroxysmal atrial fibrillation, longstanding persistent atrial fibrillation. We also do not know the status of postoperative atrial fibrillation including new onset POAF. There was a significant incomplete linking of data between the STS and the CMS database, and about 37% of patients were unable to be linked and again some patients received atrial fibrillation surgery. These patients were excluded from the analysis. Our conclusions are that preoperative atrial fibrillation is common, and about 10% of patients undergoing isolated CABG are associated with higher perioperative mortality and morbidity, excluding deep sternal wound infection. Mortality and risk of stroke and systemic embolization remains high, even adjusting for all of the comorbidities that I listed before. A higher CHADS-VASC score was associated with worse long-term outcomes, including stroke and death. Clearly the implication here is that further analysis of the comparative effectiveness of concomitant AF ablation have important implications for this very-high-risk cohort. Thank you very much. Dr Mokadam. Discussion will be led by Dr Caine.
William T. Caine. Hi. Thanks Dr Mitchell and _____ [blank tape], and I congratulate Dr Malaisrie for this excellent study. You and your colleagues have done a fantastic job, and I appreciate the advance copy of the manuscript. This study, I think, demonstrates a great use of the wealth of data that exists in the STS database, and especially the Medicare linkage that allows long-term mortality statistics to be derived, which are actually contained in the database itself. And you can tell it’s a powerful study, because I noticed in your data that the patients with preoperative AFib were 3 mm taller than the patients who didn’t have AFib, and that was strongly statistically significant, which makes me think that being tall has more disadvantages than just being uncomfortable on airplanes. Anyway, I was surprised to see that in this day and age that only two thirds of the patients in your study had preoperative atrial fibrillation, had no attempt at any procedure, or any ablation, or anything like that to treat their atrial fibrillation, and I wondered if you noticed in your data if in the later years of the study the patients, obviously in later years, if the trend to performing an ablation had grown. Dr Malaisrie. You have to remember that this data set represents all of the cases done in the United States, as you know, and probably in this room most of us would go ahead and perform surgical ablation, and the 30% is about what has been seen in previous data sets from the STS. The undertreatment of atrial fibrillation has decreased over the years. I, unfortunately, do not have that data to show you at this presentation. Dr Caine. Thank you. I also noticed that your hazard ratio is a little bit lower than some of the other published studies that have looked at this question, and I wonder is that because of the wealth or the breadth of your data and better risk adjustment? Or what would you attribute that to? Dr Malaisrie. The Medicare database is an administrative database, and you have to rely on capturing endpoints such as stroke and systemic embolization through ICD-9 codes, so it’s very possible that the rate of stroke is underreported in this certain data set. Certainly, the hard endpoint of mortality, I think, is a reliable one with stroke and systemic embolization likely underreported. Dr Caine. And just the last question. I think we’re all anxious to see if there’s some way to intervene in this process with ablation, if that will improve outcomes and look forward to the results of the data that you’re looking at now, especially the mortality cost of the stroke and systemic embolization issues that you highlighted in your presentation is really staggering. How have you at Northwestern used this data to change your practice, and how should I as a practicing cardiac surgeon implement your data into my day-to-day cardiac surgery practice?
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
9
!ACQ
Malaisrie et al
Acquired
Dr Malaisrie. I think, Bill, you and I are in centers that advocate surgical ablation if it is present at the time of cardiac surgery. I think this data set, I think should help convince the other 7% of surgeons out there that there is a high cost for perioperative or preoperative atrial fibrillation. In particular, patients with very high CHADS-VASC scores, so if you can identify a group of patients that are at risk for stroke and mortality, certainly we would want to bring their survival curve, risk of stroke, down by any way possible. We know that Maze procedure has been shown to be a very safe procedure without increasing perioperative morbidity and mortality. The downside is, of course, cost. I also look forward to the second part of our analysis when we look at the comparative data between the patients that did, in fact, have surgical ablation and that data set is pending. Dr Caine. Thank you. Congratulations again on an excellent _____.
Malaisrie et al
John D. Mitchell. I have a question for you. Great presentation, very clearly done. Are you familiar with how the STS _____ is calculated? And as a result of the data, should it be adjusted? Dr Malaisrie. I think when you actually do these by hand, you’ll notice that you do click on preoperative atrial fibrillation, and STS score and morbidity do change. Dr Mitchell. _____. Dr Malaisrie. Yeah, I think this data set has been incorporated in the current STS PROM. I think there has been another iteration of the calculator recently, and the question is whether or not it is overestimated what we see in practice. I don’t know the answer to that. Dr Mitchell. Thank you.
!ACQ 10
The Journal of Thoracic and Cardiovascular Surgery c - 2018
Acquired
!ACQ
Malaisrie et al
FIGURE E1. Flow diagram for patients undergoing isolated coronary artery bypass grafting as their first cardiovascular surgery with discharge dates from January 1, 2006, to December 31, 2013. CV, Cardiovascular; CABG, coronary artery bypass graft; IABP, intra-aortic balloon pump; PCI, percutaneous coronary intervention; STS, Society of Thoracic Surgeons; CMS, Centers for Medicare and Medicaid; AF, atrial fibrillation.
The Journal of Thoracic and Cardiovascular Surgery c Volume -, Number -
10.e1
Acquired
000
Burden of preoperative atrial fibrillation in patients undergoing coronary artery bypass grafting S. Chris Malaisrie, MD, Patrick M. McCarthy, MD, Jane Kruse, BSN, Roland Matsouaka, PhD, Adin-Cristian Andrei, PhD, Maria V. Grau-Sepulveda, MD, Daniel J. Friedman, MD, James L. Cox, MD, and J. Matthew Brennan, MD, Chicago, Ill, and Durham, NC AF before CABG is associated with higher early and late perioperative mortality and morbidity. Survival and risk of stroke or systemic embolism are worse after adjusting for comorbidities.
!ACQ The Journal of Thoracic and Cardiovascular Surgery c - 2018
Malaisrie et al