Abstract
Objective:
We sought to elucidate national practice patterns regarding anticoagulation and antiarrhythmic medication use at discharge and examine short-term patient outcomes.
Methods:
In this retrospective cohort study, we analyzed patient data from the Society of Thoracic Surgery (STS) Adult Cardiac Surgery Database from July 2011-June 2018 who underwent first-time isolated coronary artery bypass graft surgery (CABG) and developed new post-CABG atrial fibrillation (AF) without significant complication. 166,747 patients met study criteria. We examined 30-day outcomes.
Results:
166,747 patients were analyzed, and were divided into 4 groups based on discharge medications: amiodarone with or without anticoagulation, anticoagulation alone, and neither. Demographic characteristics were similar amongst the four groups. 25.7% of patients were discharged on anticoagulation with an average CHA2DS2-VASc score of 3.2 +/−1.3. Anticoagulation use at discharge was not associated with lower 30-day stroke readmissions [Adjusted Odds Ratio (AOR) 0.87, 95% CI 0.65–1.16, p=0.35]. Adjusted 30-day readmissions for major bleeding were significantly more common in anticoagulated patients (AOR 4.30, 95%CI 3.69–5.03, p<.0001). Among those discharged off anticoagulation, there was no significant difference in adjusted 30-day stroke rates based on amiodarone use at discharge (AOR 1.19, 95% CI 0.85–1.66, p=0.31).
Conclusions:
Post-CABG anticoagulation for new AF is associated with increased bleeding and no difference in stroke at 30 days. Prospective randomized studies are needed to formalize safe and efficacious short- and long-term management strategies.
Introduction
New-onset atrial fibrillation (AF) occurs in the post op period in up to 30% of coronary artery bypass grafting (CABG) patients.1,2 Despite its frequency and clinical importance, recommendations for the use of anticoagulation at hospital discharge following CABG are absent from societal guidelines and data on patient outcomes is lacking.3,4
Guideline based management of atrial fibrillation advises anticoagulation to be administered based on CHA2DS2-VASc risk factors, irrespective of rhythm status.4 The administration of antiarrhythmic drugs in patients while in AF carries a risk of chemical cardioversion and thromboembolism in the absence of concomitant anticoagulation.5 It is unclear if these basic tenets of AF related thromboembolic prophylaxis are applicable in contemporary post-CABG AF care.
We sought to elucidate national practice patterns regarding anticoagulation with and without amiodarone (the predominant post-CABG antiarrhythmic drug used)6 and examine short-term patient outcomes in order to determine current trends in management and suggest directions for further research.
Methods
Patient Population
For this retrospective cohort study, we included patients from STS data versions 2.73, 2.81 and 2.9 from July 1, 2011 to June 30, 2018 undergoing first-time elective CABG with no concurrent valve procedure who developed new post-operative AF, defined on data collection forms as post-operative AF lasting longer than one hour and/or requiring treatment.7
Because this was an evaluation of discharge management strategies, only patients discharged alive are included in this analysis. Using variables from the STS data collection forms, we excluded patients with a prior history of AF (either paroxysmal or persistent) or anticoagulation use in the previous 48 hours prior to surgery, prior use of antiarrhythmics, history of stroke, previous CABG, previous ventricular assist device, or previous valve operation. We also excluded any patients who required extra corporeal membrane oxygenation peri-procedurally, re-operation for any reason, and those who developed perioperative venous thromboembolism, tamponade, aortic dissection, stroke, cardiogenic shock, gastrointestinal bleed, or anticoagulation event (defined as bleeding, hemorrhage, and/or embolic events related to anticoagulant therapy post-operatively). Patients in an unresponsive state or in cardiogenic shock at time of procedure were also excluded, as well as patients with missing discharge medication data (Figure 1).
Figure 1. Study Patient Population.
1,075,433 patients in the STS database underwent CABG between July 1,2011 and June 30, 2018. Patients with prior history of stroke/transient ischemic attack (TIA) and prior atrial fibrillation were excluded to isolate for patients with new-onset atrial fibrillation post-CABG. Patients with significant in-hospital complications were also excluded as they would have competing risks/indications for initiation anticoagulation at discharge. Patients who died in-hospital were also excluded.
CABG = coronary artery bypass grafting
CVA = cerebrovascular accident
ECMO = extracorporeal membrane oxygenation
OR = operating room
STS = Society of Thoracic Surgery
TIA = transient ischemic attack
VAD = ventricular assist device
VTE = venous thromboembolism
Exposure Groups
The primary exposures of interest were anticoagulation and/or amiodarone prescribed at discharge from the hospital, as captured in the data collection form. Oral anticoagulants documented in the database include warfarin in addition to direct oral anticoagulants including rivaroxaban, dabigatran, apixaban, and edoxaban. Other antiarrhythmic drugs besides amiodarone were not collected in the STS database. Patients were initially analyzed based on whether or not they were discharged on an oral anticoagulant (AC) without regard to amiodarone use, and secondarily, further stratified by amiodarone (AR) use. This created 4 groups-those not prescribed anticoagulation nor amiodarone (No AR/no AC), those prescribed amiodarone without an oral anticoagulant (AR/No AC), those prescribed an oral anticoagulant without amiodarone (No AR/AC), and those prescribed both an oral anticoagulant and amiodarone (AR/AC).
Outcomes
The primary outcomes were 30-day readmissions for stroke of any type and readmission for bleeding as assessed by the STS ACSD. A secondary outcome was 30-day mortality, though, as previously mentioned, patients with peri-operative in-hospital death were excluded.
Statistical Analysis
Baseline characteristics of the study population were analyzed based on the four treatment groups discussed above. Data was summarized using frequency counts and percentages for categorical variables and means and standard deviations for continuous variables. In addition, anticoagulation and amiodarone prescription patterns were assessed as a function of hospital geography, individual hospitals, and baseline CHA2DS2-VASc scores. The database was unable to compute HASBLED scores for each patient. Missing data was imputed to “NO” or lowest risk value from binary/categorical variables consistent with approach taken in previously validated and published STS models.8,9
We compared unadjusted 30-day observed event rates by treatment group. For unadjusted analysis, outcomes were compared across study groups using chi-square tests. 30-day unadjusted outcomes were also assessed for those with very high (5–9) CHA2DS2-VASc scores, as this score meets any score threshold for anticoagulation treatment, regardless of gender.10 In addition, unadjusted outcomes were also assessed based on type of oral anticoagulant prescribed (warfarin versus direct oral anticoagulant (DOAC)).
For adjusted analysis, a generalized estimated equations method was used to account for hospital clustering. To adjust for covariates in overall readmissions, a previously used CABG operative mortality model was applied.9 This model was adjusted for assessing mortality, bleeding, and stroke readmissions. For mortality and bleeding, model differences included: age was used as one linear continuous variable, glomerular filtration rate (GFR) was used instead of creatinine, any prior percutaneous coronary intervention was assessed regardless of time performed, and prior myocardial infarction within 21 days was included. For stroke readmissions, there were no interaction terms and no reoperations were included. Logistic regression was used to compute odds ratios (OR) and 95% CI. For missing data, lowest risk missing values were imputed for binary variables in model-an approach taken in previously validated and published STS models. Adjusted analysis was also performed based on type of oral anticoagulant prescribed. This analysis was not adjusted for multiple comparisons. When testing 20 independent hypotheses, the probability of observing at least one p-value less than 0.05 by chance alone is 64%. This needs to be taken into consideration when interpreting the results.
Results
Overall Patient Population
Between July 2011 and June 2018, 832,958 patients without prior AF, stroke, or prior anticoagulant use underwent isolated first-time elective CABG and 198,013 (23.8%) developed new post-operative AF. Of these patients a total of 166,747 patients met study criteria (Figure 1). Among patients who met inclusion criteria, 25.7 % were discharged on anticoagulation. Within this anticoagulated group, 79.8% received concomitant antiarrhythmic therapy with amiodarone and 20.2% received only anticoagulation without concomitant amiodarone. Within the 74.3% of the population who were discharged without anticoagulation, 76.5% received amiodarone and 23.5% received neither anticoagulation nor amiodarone.
Patient Characteristics by Treatment Group
Baseline characteristics were generally similar across treatment groups (Table 1a).
Table 1.
Patient demographics according to patient subgroup.
| No AR*/No AC† N= 29,150 |
AR/No AC N= 94,755 |
No AR/AC N = 8,672 |
AR/AC N = 34,170 |
|
|---|---|---|---|---|
| Mean Age (SD ‡) | 68.1 (9.4) | 67.7 (9.1) | 70.5 (8.8) | 69.4 (8.6) |
| Female (%) | 7260 (24.9) | 19099 (20.2) | 1993 (23.0) | 6680 (19.6) |
| Caucasian (%) | 23630 (81.1) | 79813 (84.2) | 7420 (85.6) | 29741 (87.0) |
| Medicare (%) | 7850 (26.9) | 24868 (26.2) | 2650 (30.6) | 10032 (29.4) |
| Mean BMI§ (SD) | 29.9 (6.0) | 30.1 (5.9) | 30.7 (11.7) | 30.8 (10.6) |
| Diabetes-Insulin (%) | 4897 (16.8) | 13132 (13.9) | 1501 (17.3) | 4988 (14.6) |
| Diabetes-No Insulin (%) | 8868 (30.4) | 28096 (29.7) | 2802 (32.3) | 10460 (30.6) |
| Hypertension (%) | 26066 (89.4) | 84328 (89.0) | 7990 (92.1) | 30994 (90.7) |
| Severe Chronic Lung Disease (%) | 1253 (4.3) | 3749 (4.0) | 510 (5.9) | 1661 (4.9) |
| Dialysis (%) | 967 (3.3) | 1989 (2.1) | 271 (3.1) | 741 (2.2) |
| Mean pre-operative Creatinine (SD) | 1.22 (1.09) | 1.14 (0.90) | 1.23 (1.01) | 1.15 (0.88) |
| Dyslipidemia (%) | 25888 (88.8) | 84063 (88.7) | 7773 (89.6) | 30652 (89.7) |
| Current Smoking (%) | 5361 (18.4) | 17818 (18.8) | 1326 (15.3) | 5571 (16.3) |
| Peripheral Vascular Disease (%) | 4103 (14.1) | 11932 (12.6) | 1388 (16.0) | 4919 (14.4) |
| Previous Carotid Surgery (%) | 899 (3.1) | 2559 (2.7) | 305 (3.5) | 1058 (3.1) |
| Cancer within 5 year (%) | 1431 (4.9) | 4713 (5.0) | 500 (5.8) | 1827 (5.4) |
| Liver Disease (%) | 907 (3.1) | 2317 (2.5) | 242 (2.8) | 754 (2.2) |
| Mean CHA2DS2VASc Score (SD) | 3.3 | 3.1 | 3.6 | 3.3 |
| Pre-op ADP‖ Inhibitor within 5 days (%) | 3084 (10.6) | 9009 (9.5) | 862 (9.9) | 3028 (8.9) |
| Pre-op Beta blocker (%) | 25881 (88.8) | 85688 (90.4) | 7721 (89.0) | 31155 (91.2) |
| Pre-op ACE# inhibitor-ARB** within 48 hours (%) | 12746 (43.7) | 42205 (44.5) | 3786 (43.7) | 15280 (44.7) |
| Myocardial Infarction within 1–7 days (%) | 7115 (24.4) | 22943 (24.2) | 2141 (24.7) | 8062 (23.6) |
| Pre-op IABP†† (%) | 1,471 (5.1) | 4,458 (4.7) | 437 (5.0) | 1,730 (5.1) |
Regional variability existed with respect to prescribing pattern, though the AR/no AC group was the largest in all regions of the country. Patients in the Northeast received more anticoagulation at discharge with or without amiodarone (34.1% of patients discharged with anticoagulation in the Northeast vs 23.0% in the South, 27.5% in the Midwest, 21.5% in the West (p<.0001)) while amiodarone prescribing was lowest in the Northeast (71.0% in Northeast vs 78.0% in South, 79.1% in Midwest, and 80.3% in the West (P<.0001)) (Table 1b). An individual hospital-based histogram is shown in Figure 2.
Table 1b.
Amiodarone and anticoagulation use based on geographic location of hospital (p<.0001)
| Hospital Region | West N=27538 |
Midwest N=41500 |
South N=68387 |
Northeast N=27659 |
Canada N = 1663 |
|---|---|---|---|---|---|
| No AR*/No AC† (%) | 4423 (16.1) | 6682 (16.1) | 12104 (17.7) | 5425 (19.6) | 516 (31.0) |
| AR/No AC (%) | 17209 (62.5) | 23407 (56.4) | 40551 (59.3) | 12816 (46.3) | 772 (46.4) |
| No AR/AC (%) | 1016 (3.7) | 2012 (4.9) | 2950 (4.3) | 2577 (9.3) | 117 (7.0) |
| AR/AC (%) | 4890 (17.8) | 9399 (22.7) | 12782 (18.7) | 6841 (24.7) | 258 (15.5) |
Figure 2. Hospital-based histogram demonstrating amiodarone and anticoagulation prescribing patterns.
Each vertical line represents the practice pattern at an individual hospital with respect to discharge medications, where the Y axis represents the percentage of each discharge regimen: Blue = AR/No AC, Orange = No AR/AC, Black = No AR/AC, Yellow = AR/AC. Hospitals on the left of the histogram were more likely to prescribe AR/No AC (Blue) compared to hospitals on the right of the histogram. Despite all regions of the country most commonly withholding anticoagulation and prescribing amiodarone (AR/No AC, Blue), there is substantial variability of prescribing patterns between individual hospitals.
AC = anticoagulation
AR = amiodarone
There was a trend in prescription patterns based on pre-operative CHA2DS2-VASc score. Patients with higher scores were more likely to receive oral anticoagulation, though were not more likely to receive amiodarone [CHA2DS2-VASC score 0 with 16.4% anticoagulation and 79.3% amiodarone use versus CHA2DS2-VASC score 5–9 with 29.9% anticoagulation and 73.5% amiodarone use (p<.0001)] (Table 1c).
Table 1c.
Amiodarone and anticoagulation use based on pre-operative CHA2DS2VASc score (p<.0001)
| CHADS-VASC Score | CHADS-VASC 0 N=1927 |
CHADS-VASC 1 N=13322 |
CHADS-VASC 2–4 N=124127 |
CHADS-VASC 5–9 N=27371 |
|---|---|---|---|---|
| No AR*/No AC† (%) | 348 (18.1) | 2324 (17.4) | 21292 (17.2) | 5186 (19.0) |
| AR/No AC (%) | 1263 (65.5) | 8299 (62.3) | 71185 (57.4) | 14008 (51.2) |
| No AR/AC (%) | 51 (2.7) | 429 (3.2) | 6136 (4.9) | 2056 (7.5) |
| AR/AC (%) | 265 (13.8) | 2270 (17.0) | 25514 (20.6) | 6121 (22.4) |
AC= anticoagulation use
ACE= angiotensin converting enzyme
ADP= adenosine disphosphate
AR= amiodarone use
ARB= angiotensin receptor blocker
BMI= body mass index
IABP= Intra-aortic balloon pump
SD= standard deviation
Unadjusted Outcomes Based on Anticoagulation and Amiodarone Use
Overall rates of 30-day post-discharge mortality (excluding peri-operative in-hospital death), stroke readmission, and bleeding readmission were ≤1% regardless of anticoagulation or amiodarone use (Table 2a).
Table 2a-.
Unadjusted 30-day outcomes for patients based on anticoagulation and amiodarone prescription.
| Cohort | 30-Day Stroke Readmission | 30-Day Bleeding Readmission | 30-Day Mortality | |||
|---|---|---|---|---|---|---|
| Number Events (%) | P-value | Number Events (%) | P-value | Number Events (%) | P-value | |
| No AC* [N=123905] | 210 (0.17) | 0.571 | 281 (0.23) | <0.0001 | 502 (0.41) | 0.0002 |
| AC [N=42842] | 67 (0.16) | 418 (0.98) | 238 (0.56) | |||
| No AR†/No AC [N=29150] | 45 (0.15) | 0.0024 | 73 (0.25) | <0.0001 | 139 (0.48) | 0.0002 |
| AR/No AC [N=94755] | 165 (0.17) | 208 (0.22) | 363 (0.38) | |||
| No AR/AC [N=8672] | 26 (0.3) | 70 (0.81) | 51 (0.59) | |||
| AR/AC [N=34170] | 41 (0.12) | 348 (1.02) | 187 (0.55) | |||
Overall readmission rates are listed in Supplementary Table 3.
Unadjusted 30-day readmissions for bleeding were substantially higher in patients discharged on anticoagulation (p<.0001). There was no statistically significant difference in 30-day stroke readmissions between those patients who did and did not receive anticoagulation (p=0.571) (Table 2a). There was an increased 30-day mortality seen in those patients discharged on anticoagulation compared to those who were not (p=.0002).
When evaluating patients with high CHA2DS2-VASc scores (5–9) only, 30-day bleeding readmissions remained higher among anticoagulated patients (<.0001). There was a trend toward decreased 30-day stroke readmissions in anticoagulated patients in this higher risk patient population (p=0.05) (Table 2b).
Table 2b-.
Unadjusted 30-day outcomes for patients with very high CHA2DS2VASc scores (5–9).
| Cohort | 30-Day Stroke Readmission | 30-Day Bleeding Readmission | 30-Day Mortality | |||
|---|---|---|---|---|---|---|
| Number Events (%) | P-value | Number Events (%) | P-value | Number Events (%) | P-value | |
| No AR/No AC [N=5186] | 11 (0.21) | 0.05 | 15 (0.29) | <0.0001 | 60 (1.16) | 0.03 |
| AR/No AC [N=14008] | 46 (0.33) | 27 (0.19) | 102 (0.73) | |||
| No AR/AC [N=2056] | 9 (0.44) | 11 (0.54) | 15 (0.73) | |||
| AR/AC [N=6121] | 9 (0.15) | 64 (1.05) | 56 (0.91) | |||
AC= anticoagulation use
AR= amiodarone use
Among patients in the study cohort who were discharged on an oral anticoagulant (n=42,841), 36.3% (n=15,539) were discharged on a DOAC.
Adjusted Analysis
Adjusted 30-day readmissions for major bleeding were significantly more common in patients discharged on anticoagulation (AOR 4.30, 95%CI 3.69–5.03, p<.0001). However, among those anticoagulated, there was no significance difference in bleeding based on amiodarone use (AOR 1.26, 95%CI 0.98–1.63, p=0.75) (Table 3).
Table 3-.
Adjusted 30-day outcomes for patients based on anticoagulation and amiodarone prescription
| Cohort | 30-Day Stroke Readmission | 30-Day Bleeding Readmission | 30-Day Mortality | |||
|---|---|---|---|---|---|---|
| AOR* (95%CI) | P-value | AOR (95%CI) | P-value | AOR (95%CI) | P-value | |
| No AC† | REF | 0.35 | REF | <0.0001 | REF | 0.024 |
| AC | .87 (0.65–1.16) |
4.30 (3.69–5.02) |
1.20 (1.02–1.40) | |||
| No AR‡/No AC | REF | NA | REF | NA | REF | NA |
| AR/No AC | 1.19 (0.85–1.66) | 0.31 | 0.87 (0.67–1.15) | 0.34 | 0.87 (0.71–1.07) | 0.19 |
| No AR/AC | 1.79 (1.10–2.91) |
0.02 | 3.23 (2.31–4.50) | <.0001 | 1.03 (0.74–1.43) | 0.87 |
| AR/AC | 0.78 (0.50–1.20) | 0.25 | 4.07 (3.17–5.22) | <.0001 | 1.09 (0.87–1.37) | 0.43 |
| AR/AC vs AR/No AC | 0.65 (0.46–0.92) | 0.015 | 4.65 (3.89–5.56) | <.0001 | 1.26 (1.05–1.50) | 0.014 |
| AR/AC vs No AR/AC | 0.43 (0.27–0.70) | 0.0007 | 1.26 (0.98–1.63) | 0.75 | 1.06 (0.77–1.47) | 0.70 |
Of note: Please refer to Supplemental Table 3 for overall readmission rates
AC= anticoagulation use
AOR=Adjusted odds ratio
AR=amiodarone use
After adjustment, there was no difference in 30-day stroke readmissions based on anticoagulation use (AOR 0.87, 95% CI 0.65–1.16, p=0.35) overall, though anticoagulation was associated with lower stroke rates when analyzing only patients discharged on amiodarone (AOR 0.65, 95% CI 0.46–0.92, p =.02) (Table 3). Among all patients discharged on anticoagulation, there was an association between amiodarone use and lower stroke risk (AOR 0.43, p<.001). Among those discharged off anticoagulation, there was no significant difference in adjusted 30-day stroke rates based on amiodarone use at discharge (AOR 1.19, 95% CI 0.85–1.66, p=0.31).
There was a statistically significant increase in mortality for anticoagulation use in the adjusted analysis (AOR 1.20, 95% CI 1.02–1.40, p=.02). When evaluating type of anticoagulant used, this mortality signal appears to stem from DOAC use (AOR 1.45, 95% CI 1.17–1.80, p=0.0007), while adjusted readmission for stroke and bleeding were similar based on type of oral anticoagulant used (Table 4).
Table 4-.
Adjusted 30-day outcomes for anticoagulated patients based on type of oral anticoagulant. (Total N = 42,841, Total on DOAC*= 15,539 (36.3%)). Reference group is patients discharged off anticoagulation.
| In-Hospital Outcomes | Variable | Adjusted OR | Lower 95% CI | Upper 95% CI | P-value |
|---|---|---|---|---|---|
| Operative Mortality | Warfarin | 1.05 | 0.87 | 1.26 | 0.6162 |
| DOAC | 1.45 | 1.17 | 1.80 | 0.0007 | |
| Any Readmission | Warfarin | 1.18 | 1.12 | 1.24 | <.0001 |
| DOAC | 1.23 | 1.16 | 1.29 | <.0001 | |
| Readmission for Stroke | Warfarin | 0.90 | 0.65 | 1.25 | 0.5250 |
| DOAC | 0.82 | 0.53 | 1.27 | 0.3761 | |
| Readmission for Bleeding | Warfarin | 4.54 | 3.82 | 5.39 | <.0001 |
| DOAC | 3.89 | 3.17 | 4.78 | <.0001 |
DOAC- Direct oral anticoagulant
We found no difference in clinical outcomes when assessing patients only based on the use of amiodarone (Supplemental Table 4).
Discussion
This retrospective cohort study is the first to evaluate nationwide practice patterns among patients who underwent first-time elective CABG and developed new post-operative AF. We found a majority of patients are not prescribed anticoagulation at discharge in this setting, while a high percentage are prescribed amiodarone. Adjusted 30-day stroke readmissions and mortality were similar regardless of anticoagulation status at discharge. However, adjusted 30-day readmissions for major bleeding were significantly more common in anticoagulated patients (AOR 4.30, 95%CI 3.69–5.03, p<.0001). These results shed new light on national practice patterns for a common scenario encountered in cardiac surgical patients, and raise questions regarding the optimal medical regimen for new AF in the immediately post-operative period.
We chose this homogeneous cohort of isolated elective CABG patients and excluded those with concomitant valvular surgery to prevent ambiguity regarding the indication for oral anticoagulation. Our analysis was restricted to the early post-operative period (first 30 days) and excluded patients who developed significant peri-operative complications to isolate a population of patients for whom anticoagulation could reasonably be considered.
Both medical (American Heart Association/American College of Cardiology/Heart Rhythm Society, European Society of Cardiology) and surgical (American Academy of Thoracic Surgeons) societies currently recommend anticoagulation for post-surgical AF, particularly when deemed clinically significant, defined by the requirement of rate or rhythm control agents or when the AF extends the duration of the hospitalization.3,4,11,12 Guidelines also suggest the initiation of antiarrhythmic therapy to maintain sinus rhythm in post-operative patients who are considered high risk for bleeding complications with anticoagulation. This latter recommendation is based upon the as yet unsubstantiated premise that rhythm control reduces the risk of stroke. In practice, multi-disciplinary post-operative teams tend to avoid the bleeding risks of anticoagulation, and utilize anti-arrhythmic drug therapy to limit AF burden and shorten hospital length of stay.13–16
Our analysis supports this approach as only 25.7% of those with post-operative AF were discharged on anticoagulation despite a mean CHA2DS2-VASc score of 3.2. In addition, over three quarters of patients with post-CABG AF were discharged on amiodarone and 56.8% of the total population received amiodarone without concomitant anticoagulation. This rate of anticoagulation was consistent with rates reported in the sub-analysis of post-CABG AF patients in the PREVENT-IV trial Outcomes analysis based on anticoagulation use was not performed in this study.
Though some regional variability was apparent, withholding anticoagulation while prescribing amiodarone was by far the most common practice in all regions. This was also the most common practice even among patients with very high risk of a thromboembolic event (CHA2DS2-VASc 5–9).
Anticoagulation for post-CABG AF was associated with a greater than four-fold increase in the risk of major bleeding in our study. Although amiodarone carries a risk of excessive INR elevation when given with warfarin, our analysis showed no increased bleeding risk associated with this combination.
Stroke risk associated with post-op AF may be lower with cardiac surgery compared to non-cardiac surgery.2,17 Also, a separate retrospective analysis from a Danish nationwide registry demonstrated lower thromboembolism rates in post-CABF AF compared with a non-valvular AF cohort unassociated with surgery.2 Unlike these studies, ours focused only on the first 30 days after CABG. We found no association between anticoagulation use and stroke risk during this early post-operative period.
Our analysis did not allow patient level determination of the heart rhythm status while on amiodarone. Therefore, we cannot determine if amiodarone was used to prevent AF recurrence or started while the patient was in AF. The latter circumstance would theoretically place the un-anticoagulated patient at higher risk of stroke should chemical cardioversion occur.5,18 Given the frequent use of amiodarone in un-anticoagulated patients in our study, it is likely that many were exposed to this potential risk. Notably, among those discharged off of anticoagulation, amiodarone use was not associated with an increase in adjusted 30-day stroke rate (AOR 1.19, 95% CI 0.85–1.66, p=0.31). However, the combination of amiodarone and anticoagulation was associated with a lower stroke readmission risk (AOR 0.65, 95% CI 0.46–0.92, p =.015) when compared to those on amiodarone alone. Prospective research is necessary to evaluate the appropriate use of amiodarone in relation to anticoagulation in this setting.
The increased mortality for anticoagulation use seen in the adjusted analysis is concerning. However, this may be tied to the higher hospital readmission rates in this population and unmeasured comorbidity differences between the anticoagulated and un-anticoagulated populations.
In our analysis, over a third of the anticoagulated patients received DOACs. Data exists for an increased risk of bleeding complications with DOACs in the immediate peri-cardiac surgery period but is lacking for the period studied in our analysis.19 In our population there was no difference in adjusted 30 day risk of bleeding or stroke associated with DOAC use compared with warfarin. However, the adjusted mortality rates were significantly higher in patients discharged with DOACs compared with warfarin. This is unlikely explained by the faster onset of DOAC medications compared to warfarin because bleeding readmissions were similar among oral anticoagulants. As DOAC use is now the preferred anticoagulation strategy in atrial fibrillation given the complexities of warfarin titration, more data is needed regarding their safety in post-CABG AF.20,21
Overall, 30-day event rates for bleeding, stroke, and mortality were low following an uncomplicated CABG. Anticoagulation use was not associated with a statistically lower stroke rate at 30 days, but there was a clear association of bleeding risk among all patients who were discharged with anticoagulation. Multiple retrospective studies have suggested that higher risk patients with post-CABG AF may need systemic anticoagulation in the long term. Our results raise the possibility that initiation of anticoagulation may be best delayed for several weeks to diminish the risk of post-operative bleeding. This will be hopefully be addressed soon as discussions for a randomized trial for anticoagulation for post-CABG AF are underway.
Limitations
This study is a retrospective, descriptive analysis and thus we cannot draw any definitive conclusions regarding causality. Similarly, it is possible that unknown confounders contributed to study results. The STS data collection forms define AF as lasting longer than an hour or requiring treatment, so shorter episodes of AF may not have been coded, resulting in an underestimate of the true incidence of post-CABG AF. Further, we do not know the burden of AF nor patients’ discharge heart rhythm. This information likely affected practice patterns, particularly amiodarone use. With this study, we simply challenge the guideline-based standard of care, which is to anti-coagulate patients with clinically significant AF regardless of rhythm status based on CHADS2VASc score. Based on our results, we feel that uniform application of the CHADS2VASc score may not always be warranted.
We excluded patients who developed post-operative cardiac tamponade. However, a small percentage of patients may have developed tamponade as a result of anticoagulation initiation. Patient data was only collected for 30 days after discharge and outcomes beyond 30 days are unknown. The STS database does not include other antiarrhythmic drugs such as sotalol or dofetilide which may have been used in this population. It is likely however that these agents are used in small proportion compared with amiodarone.6 Additionally, with respect to discharge medications, the STS database cannot determine which patients filled discharge prescriptions or were compliant with their discharge regimen. We also could not ascertain which patients, if any, had a left atrial appendage excision at the time of surgery. We anticipate this happened infrequently as our cohort by design had no history of pre-operative AF. Finally, this dataset does not specify the type of stroke on readmission (ischemic vs hemorrhagic), so the mechanism of stroke remains unknown.
Conclusions
The predominant national practice for post-CABG AF is to withhold anticoagulation and prescribe amiodarone. This strategy of withholding anticoagulation in the four weeks post-CABG, which challenges the recommendations of both Cardiac and Cardiac Surgery Guidelines, may be warranted given significantly higher rates of readmission for bleeding among patients discharged on anticoagulation (Figure 3). Prospective studies are required to better define the role of anticoagulation in the early post CABG period.
Figure 3. Graphical abstract of study methods and results.
166,747 patients in the STS database with new atrial fibrillation after first time isolated CABG met the study criteria. Of these, the majority were discharged on amiodarone without anticoagulation, despite guidelines. There was no significant difference in adjusted 30-day stroke readmission in patients discharged on anticoagulation, but anticoagulation was associated with significantly higher risk for 30-day bleeding readmission.
CABG = Coronary Artery Bypass Graft
STS = Society of Thoracic Surgery
AF = Atrial Fibrillation
AC = Anticoagulation use
AR = Amiodarone use
Supplementary Material
Central Message.
The predominant national practice of discharging patients with new post-CABG AF on amiodarone without anticoagulation challenges guidelines, but may be warranted given postoperative bleeding risk.
Perspective Statement.
Post-CABG AF is common, and is associated with increased morbidity and mortality. However, management of AF following CABG is not standardized. Nationwide, the majority of post-CABG AF patients were discharged off anticoagulation and on amiodarone, contrary to guidelines, but with no increased stroke risk. Prospective randomized studies are needed to formalize management strategies.
Acknowledgements:
The data for this research were provided by The Society of Thoracic Surgeons’ National Database Access and Publications Research Program. We would also like to thank Linda Valsdottir for her help with creation of the Visual Abstract.
Glossary of Abbreviations
- AC
anticoagulation use
- ACSD
Adult Cardiac Surgery Database
- AF
atrial fibrillation
- AOR
adjusted odds ratio
- AR
amiodarone use
- CABG
coronary artery bypass grafting
- DOAC
direct oral anticoagulant
- OR
odds ratio
- STS
Society for Thoracic Surgery
Footnotes
None of the authors have any conflicts of interest or sources of funding that relate to this study.
Justification for >7 authors: Collaboration between General Cardiology, Electrophysiology, and Cardiothoracic Surgery, as well as statisticians from Duke Clinical Research Institute
References:
- 1.Echahidi N, Pibarot P, O’Hara G, Mathieu P. Mechanisms, prevention, and treatment of atrial fibrillation after cardiac surgery. J Am Coll Cardiol 2008;51:793–801. [DOI] [PubMed] [Google Scholar]
- 2.Butt JH, Xian Y, Peterson ED, et al. Long-term Thromboembolic Risk in Patients With Postoperative Atrial Fibrillation After Coronary Artery Bypass Graft Surgery and Patients With Nonvalvular Atrial Fibrillation. JAMA Cardiol 2018;3:417–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Frendl G, Sodickson AC, Chung MK, et al. 2014 AATS guidelines for the prevention and management of perioperative atrial fibrillation and flutter for thoracic surgical procedures. Executive summary. J Thorac Cardiovasc Surg 2014;148:772–91. [DOI] [PubMed] [Google Scholar]
- 4.January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the Heart Rhythm Society. Circulation 2014;130:e199–267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Verhaert D, Puwanant S, Gillinov AM, Klein AL. Atrial fibrillation after open heart surgery: how safe is early conversion without anticoagulation? J Am Soc Echocardiogr 2009;22:212 e1–3. [DOI] [PubMed] [Google Scholar]
- 6.Al-Khatib SM, Hafley G, Harrington RA, et al. Patterns of management of atrial fibrillation complicating coronary artery bypass grafting: Results from the PRoject of Ex-vivo Vein graft ENgineering via Transfection IV (PREVENT-IV) Trial. Am Heart J 2009;158:792–8. [DOI] [PubMed] [Google Scholar]
- 7.Adult Cardiac Surgery Database Collection. 2019. (Accessed May 13, 2019, at https://www.sts.org/registries-research-center/sts-national-database/adult-cardiac-surgery-database/data-collection.)
- 8.Shahian DM, O’Brien SM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 3--valve plus coronary artery bypass grafting surgery. Ann Thorac Surg 2009;88:S43–62. [DOI] [PubMed] [Google Scholar]
- 9.Shahian DM, O’Brien SM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 1--coronary artery bypass grafting surgery. Ann Thorac Surg 2009;88:S2–22. [DOI] [PubMed] [Google Scholar]
- 10.Potpara TS, Lip GYH, Blomstrom-Lundqvist C, Chiang CE, Camm AJ. Viewpoint: Stroke Prevention in Recent Guidelines for the Management of Patients with Atrial Fibrillation: An Appraisal. Am J Med 2017;130:773–9. [DOI] [PubMed] [Google Scholar]
- 11.Maisel WH, Rawn JD, Stevenson WG. Atrial fibrillation after cardiac surgery. Ann Intern Med 2001;135:1061–73. [DOI] [PubMed] [Google Scholar]
- 12.Kirchhof P, Benussi S, Kotecha D, et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Europace 2016;18:1609–78. [DOI] [PubMed] [Google Scholar]
- 13.Gillinov AM, Bagiella E, Moskowitz AJ, et al. Rate Control versus Rhythm Control for Atrial Fibrillation after Cardiac Surgery. N Engl J Med 2016;374:1911–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Mitchell LB, Exner DV, Wyse DG, et al. Prophylactic Oral Amiodarone for the Prevention of Arrhythmias that Begin Early After Revascularization, Valve Replacement, or Repair: PAPABEAR: a randomized controlled trial. JAMA 2005;294:3093–100. [DOI] [PubMed] [Google Scholar]
- 15.Halonen J, Loponen P, Jarvinen O, et al. Metoprolol versus amiodarone in the prevention of atrial fibrillation after cardiac surgery: a randomized trial. Ann Intern Med 2010;153:703–9. [DOI] [PubMed] [Google Scholar]
- 16.Samuels LE, Holmes EC, Samuels FL. Selective use of amiodarone and early cardioversion for postoperative atrial fibrillation. Ann Thorac Surg 2005;79:113–6. [DOI] [PubMed] [Google Scholar]
- 17.Butt JH, Olesen JB, Havers-Borgersen E, et al. Risk of Thromboembolism Associated With Atrial Fibrillation Following Noncardiac Surgery. J Am Coll Cardiol 2018;72:2027–36. [DOI] [PubMed] [Google Scholar]
- 18.Chen WC, Chen WC, Chen CY, et al. Amiodarone use is associated with increased risk of stroke in patients with nonvalvular atrial fibrillation: a nationwide population-based cohort study. Medicine (Baltimore) 2015;94:e849. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lisle TC, Barrett KM, Gazoni LM, et al. Timing of stroke after cardiopulmonary bypass determines mortality. Ann Thorac Surg 2008;85:1556–62; discussion 62–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.January CT, Wann LS, Calkins H, et al. 2019 AHA/ACC/HRS Focused Update of the 2014 AHA/ACC/HRS Guideline for the Management of Patients With Atrial Fibrillation: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Rhythm Society. Heart Rhythm 2019. [DOI] [PubMed]
- 21.Brennan JM, Alexander KP, Wallace A, et al. Patterns of anticoagulation following bioprosthetic valve implantation: observations from ANSWER. J Heart Valve Dis 2012;21:78–87. [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.