Accepted Manuscript Increased late mortality following coronary artery bypass surgery when complicated by isolated new onset atrial fibrillation: a comprehensive propensity-matched analysis Laila Al-Shaar , MS,MPH Thomas A. Schwann , MD Ameer Kabour , MD Robert H. Habib , PhD PII:
S0022-5223(14)00556-X
DOI:
10.1016/j.jtcvs.2014.05.020
Reference:
YMTC 8606
To appear in:
The Journal of Thoracic and Cardiovascular Surgery
Received Date: 20 November 2013 Revised Date:
25 April 2014
Accepted Date: 8 May 2014
Please cite this article as: Al-Shaar L, Schwann TA, Kabour A, Habib RH, Increased late mortality following coronary artery bypass surgery when complicated by isolated new onset atrial fibrillation: a comprehensive propensity-matched analysis, The Journal of Thoracic and Cardiovascular Surgery (2014), doi: 10.1016/j.jtcvs.2014.05.020. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
Increased late mortality following coronary artery bypass surgery when complicated by
2
isolated new onset atrial fibrillation: a comprehensive propensity-matched analysis
3
Brief Title:
Postoperative Atrial Fibrillation and CABG Survival
RI PT
4 5
Laila Al-Shaar, MS,MPH*, Thomas A. Schwann, MD‡, Ameer Kabour, MD§, and Robert
7
H. Habib, PhD*,†,¶
SC
6
8 9
*
10
†
11
American University of Beirut, Beirut, Lebanon.
12
‡
Department of Surgery, University of Toledo - College of Medicine, Toledo, Ohio, USA.
13
§
Mercy Saint Vincent Medical Center, Toledo, Ohio, USA.
M AN U
TE D
Funding Research supported by institutional and departmental funds. Author LA was supported by intramural grant funding to the Vascular Medicine Program and National Institutes of Health Fogarty Grant (NIH-FIC-D43TW009118) awarded to The Scholars in HeAlth Research Program at American University of Beirut.
EP
20 21 22 23 24 25 26 27 28 29 30 31 32 33
Outcomes Research Unit and Department of Internal Medicine,
Conflict of interests: None declared ¶
AC C
14 15 16 17 18 19
Vascular Medicine Program, American University of Beirut, Beirut, Lebanon.
Address Correspondence and Requests for Single Reprints to: Robert H. Habib, PhD Director, Outcomes Research Unit American University of Beirut-Medical Center PO BOX: 11-0236, Riad El Solh 1107 2020, Beirut, Lebanon Tel:+961-1-374444 ext.5446 E-mail: [email protected] Word Count: 3,495 words
1
ACCEPTED MANUSCRIPT
1 2 3 4 5
Abstract
6
after coronary artery bypass grafting (CABG) is confounded by the frequent concomitant serious
7
complications that co-occur with POAF. We aimed to define the magnitude and time-
8
dependence of the effect of isolated POAF on late survival after uncomplicated CABG in a
9
manner that comprehensively accounts for comorbidity and perioperative confounding factors.
10
Methods: Non-salvage CABG patients with no history of AF, no concomitant aortic/valvular
11
surgery, and no perioperative complications other than POAF were studied (N=6305). Patients
12
were divided to AF (N=1211, 68 yrs, 72% male) and No-AF (N=5094, 63 yrs, 70% male)
13
groups. Propensity matching was done based on 55 patient variables including coronary grafts,
14
completeness of revascularization and transfusion data. The AF-effect was quantified using time-
15
segmented hazard ratios (HRt) by Cox regression.
16
Results: Single (1-to-1), double (1-to-2) and triple (1-to-3) propensity matching of AF to No-AF
17
was achieved in 1196, 993 and 719 cases, respectively. AF showed significantly worse, yet time-
18
varying, 0-18 year survival: (0-1 yrs: HR=1.18 (0.77-1.81), 1-6 yrs: HR=1.37 (1.12-1.67), 6-17
19
yrs: 1.25 (1.05-1.49)).
20
Conclusions: Isolated POAF is associated with a time-varying increase in mortality after CABG
21
surgery. Given these findings and the high incidence of POAF, efforts to reduce POAF should be
22
pursued to potentially improve resource utilization, morbidity and mortality.
RI PT
SC
M AN U
TE D
EP
AC C
23
Objectives: The association of new-onset postoperative atrial fibrillation (POAF) and late death
24 25 26 2
ACCEPTED MANUSCRIPT
1
Ultra mini-abstract
3
This study investigated long term outcomes of new onset atrial fibrillation (POAF) following
4
coronary artery bypass surgery (CABG). Using multiple propensity matching methods, we found
5
that POAF increases long term mortality over an 18 year postoperative follow up period.
RI PT
2
AC C
EP
TE D
M AN U
SC
6
3
ACCEPTED MANUSCRIPT
1 2
Introduction New onset postoperative atrial fibrillation (POAF) is a frequent complication of cardiac (10%-65%) surgery1-6, and it is associated with increased morbidity, in-hospital mortality, and
4
greater resource utilization.1-10 Imbalance of the autonomic nervous system, enhanced
5
inflammation, ischemia and predisposing anatomic substrate has been implicated in PAOF,
6
however, the exact pathophysiology of POAF is poorly understood. The incidence of POAF in
7
coronary bypass grafting (CABG: 15%-40%) is reportedly predicted by patient demographics
8
including older age, lower ejection fraction, hypertension, heart failure, prolonged
9
cardiopulmonary bypass, enlarged left atrium, perioperative medications and enhanced
11
SC
inflammatory and oxidative stress.7-10
M AN U
10
RI PT
3
A number of recent studies have extended the adverse effects of POAF after CABG to include significantly worse late mortality.2, 9, 11-16 If true, this finding is of substantial clinical
13
importance given the high incidence of POAF, yet it is not clear what is the specific mechanism
14
by which an ostensibly limited perioperative arrhythmia, such as POAF, affects long term
15
survival. The available evidence on this POAF - late mortality association is limited by non-
16
uniformity of analyses and findings and/or incomplete accounting for known determinants of late
17
mortality after CABG. No studies, thus far, account for a) extent of arterial grafting and b)
18
completeness of coronary revascularization.17-22 Furthermore, homologous blood product
19
transfusion, which is notably more frequent in patients with higher likelihood of POAF, and has
20
been identified as a predictor of late mortality has also been largely ignored in studies of the
21
effects of POAF.23-25 In addition, POAF is frequently associated with other serious postoperative
22
complications and, consequently the interplay between these non-arrhythmic complications and
23
POAF may confound the independent effects of these conditions on operative and possibly late
AC C
EP
TE D
12
4
ACCEPTED MANUSCRIPT
death.2, 8, 9, 11 Separating the late effects of POAF and concurrent complications is difficult and
2
studies, thus far, have either ignored this 13, 26 or have relied on multivariate Cox regression
3
modeling to account for some but not all such complications.2, 9, 11, 14, 27 Lastly, the potentially
4
time-varying POAF effect on late mortality, its duration and the time period of highest risk, have
5
not been described.
6
RI PT
1
Within this construct, we reasoned that if POAF independently leads to increased late CABG mortality, then it should do so even in 1) the absence of other non-arrhythmic
8
postoperative complications, and 2) after more comprehensive adjustment for other known pre,
9
intra, and post-operative predictors of late survival. Accordingly, we investigated the POAF late
M AN U
SC
7
10
CABG survival association in a manner that avoids the above-mentioned limitations by more
11
comprehensive accounting for patient variables, operative and grafting details, transfusion status
12
and most importantly by excluding patients with other complications.
14
TE D
13
Methods
This investigation is a retrospective analysis of a prospectively collected Society of
16
Thoracic Surgeons (STS) Adult Cardiac Surgery National Database approved by the Mercy Saint
17
Vincent Medical Center (MSVMC, Toledo, OH; 1994-2007) Institutional Review Board. Data
18
collection was performed in compliance with the STS Database definitions and criteria and did
19
not involve additional review of hospital records or interview of patients, thus informed consent
20
requirement was waived.
22
AC C
21
EP
15
Patients. Patients undergoing isolated CABG between 1994 and 2007 were included in
23
the analysis. CABG patients with concurrent coronary or carotid reconstruction were also
24
included. Patients were excluded in case of concomitant valvular, congenital cardiac or aortic 5
ACCEPTED MANUSCRIPT
surgery, documented history of preoperative atrial fibrillation/flutter, or emergency-salvage
2
status. A total of 7,610 patients qualified for inclusion, and these were subsequently grouped as
3
follows: (1) 5,094 patients who did not develop POAF or any other postoperative complication
4
(No-AF; control group); (2) 1,211 patients with isolated POAF without other complications (AF;
5
study group); (3) 832 patients without POAF but who developed other complications (No-AF/C);
6
and (4) 473 patients who developed POAF and other complications (AF/C). To mitigate any
7
potential confounding effects of concurrent non-rhythm related perioperative complications on
8
long term survival, the subcohort of patients without any other perioperative complications aside
9
from POAF was selected as the principal focus of the investigation. Time zero was defined as
M AN U
SC
RI PT
1
time of discharge. In hospital deaths were counted as patients with complications and were
11
therefore excluded from the AF and No-AF groups. POAF was defined, as per the STS Database,
12
as the occurrence of POAF or atrial flutter requiring treatment (i.e. beta blockers, calcium
13
channel blockers, amiodarone, anticoagulation, or cardioversion) throughout hospitalization or at
14
readmission within 30 days postoperatively. Telemetry was used in 100% of patients during
15
hospitalization.
TE D
10
17
EP
16
Surgery. Our surgical techniques have been previously reported 17, 19. Cardiopulmonary bypass (96%) and aortocoronary grafting (97%) were used in a large majority of patients.
19
Evidence-based critical pathways were utilized and periodically updated in compliance with
20
emerging data documented to optimize outcomes, improve care or decrease resource utilization.
21
Treatment of POAF, if it occurred, is described in the On Line Supplement. All efforts were
22
made to discharge patients in a normal sinus rhythm. Patients discharged in rate controlled atrial
AC C
18
6
ACCEPTED MANUSCRIPT
1
fibrillation were routinely discharged on warfarin, unless contraindicated. The database is not
2
granular enough to derive the subset of POAF patients discharged in a non sinus rhythm.
3
Perioperative Medications. Aspirin and β-blockers use was consistent throughout the
RI PT
4
1994 to 2007 study period. In all patients, β-blockers were routinely administered orally until the
6
morning of surgery and in stable patients were reinstituted on postoperative day 1. Accordingly,
7
time of β-blocker withdrawal (if any) was minimized in the present series of patients. An ACE
8
inhibitor was used routinely in patients with normal renal function, including those who were
9
taking one preoperatively, who had a history of congestive heart failure or who had low ejection
M AN U
SC
5
fraction (100 mm Hg) in the
11
early phases of the study period and routinely in the later phase of the study period in compliance
12
with practice guidelines. Aspirin (325 mg via nasogastric tube) was always given to patients on
13
arrival at the Cardiovascular Intensive Care Unit and was decreased to 162 mg PO daily on
14
postoperative day 1. Amiodarone (intravenous or orally), once approved for clinical use, was
15
routinely used after onset of AF, including intra-operatively. Starting July 1999, prophylactic
16
amiodarone was used routinely in all patients (daily 200 mg PO started on postoperative day 1).
17
Prior to the availability of amiodarone, quinidine and/or procainamide in conjunction with
18
calcium channel antagonists were used to treat POAF.
20
EP
AC C
19
TE D
10
Follow-up. All-cause mortality was secured from the cardiothoracic surgery follow-up
21
and verified from scheduled searches of the US Social Security Death Index database
22
(http://ssdi.genealogy.rootsweb.com) last performed in November 2011. Follow-up was between
23
56 (minimum) and 214 (maximum) months.
7
ACCEPTED MANUSCRIPT
Propensity Score Model and Matching. AF and No-AF groups exhibited significant
2
demographic and risk factor differences (Table 1, Table-S1 of Supplement). Such differences
3
confound outcome comparisons in observational comparison groups. To minimize such
4
confounding, we used propensity score matching to derive demographics, risk factor and
5
operation matched sub-cohorts of equal size. The propensity score, or equivalently the
6
probability that a given patient develops POAF, was derived via a non-parsimonious logistic
7
multivariate regression model that considered POAF status (AF or No-AF) as the dependent
8
outcome variable. A total of 55 preoperative risk factors (Tables 1, S1), demographics and
9
operative variables were entered into the model irrespective of significance. A completeness of
10
revascularization index (CRI= #Grafts-VD) derived from the difference between the number of
11
grafts (#Grafts or distal anastomoses in the case of sequential grafts) and coronary vessel disease.
12
Time-of-surgery was entered as a continuous year of series variable [1994=1, 2007 =14] to
13
account for potential variance in reporting POAF or its treatment. Highly redundant variables
14
were avoided. The resulting propensity scores were distinctly different for isolated AF versus
15
No-AF patients [mean±SD: 0.259±0.124 vs. 0.176±0.106 respectively; p 40 kg/m2) Smoker Diabetes Insulin Dependent Renal Failure Hypercholesterolemia Hypertension Chronic Lung Disease Peripheral Vascular Disease Cerebrovascular Disease
Matched CABG patients No-AF AF (n=1196) (n=1196) Mean ± SD Mean ± SD 67.6±9.5 67.5±9.5 2.04±0.25 2.04±0.25 7.9±3.6 7.8±3.5
50±11 79±35 3.2±1 0.54±0.79
EP
Ejection Fraction (EF, %) Perfusion Time (min) Number of Grafts Revascularization Index
AC C
Variable Continuous Age (yrs) Body surface area (BSA, m2) Year of surgeryΩ
Un-matched CABG patients No-AF AF (n=5094) (n=1211) Mean ± SD Mean ± SD p-value 62.5±10.6 67.6±9.6
S0022-5223(14)00556-X
DOI:
10.1016/j.jtcvs.2014.05.020
Reference:
YMTC 8606
To appear in:
The Journal of Thoracic and Cardiovascular Surgery
Received Date: 20 November 2013 Revised Date:
25 April 2014
Accepted Date: 8 May 2014
Please cite this article as: Al-Shaar L, Schwann TA, Kabour A, Habib RH, Increased late mortality following coronary artery bypass surgery when complicated by isolated new onset atrial fibrillation: a comprehensive propensity-matched analysis, The Journal of Thoracic and Cardiovascular Surgery (2014), doi: 10.1016/j.jtcvs.2014.05.020. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
Increased late mortality following coronary artery bypass surgery when complicated by
2
isolated new onset atrial fibrillation: a comprehensive propensity-matched analysis
3
Brief Title:
Postoperative Atrial Fibrillation and CABG Survival
RI PT
4 5
Laila Al-Shaar, MS,MPH*, Thomas A. Schwann, MD‡, Ameer Kabour, MD§, and Robert
7
H. Habib, PhD*,†,¶
SC
6
8 9
*
10
†
11
American University of Beirut, Beirut, Lebanon.
12
‡
Department of Surgery, University of Toledo - College of Medicine, Toledo, Ohio, USA.
13
§
Mercy Saint Vincent Medical Center, Toledo, Ohio, USA.
M AN U
TE D
Funding Research supported by institutional and departmental funds. Author LA was supported by intramural grant funding to the Vascular Medicine Program and National Institutes of Health Fogarty Grant (NIH-FIC-D43TW009118) awarded to The Scholars in HeAlth Research Program at American University of Beirut.
EP
20 21 22 23 24 25 26 27 28 29 30 31 32 33
Outcomes Research Unit and Department of Internal Medicine,
Conflict of interests: None declared ¶
AC C
14 15 16 17 18 19
Vascular Medicine Program, American University of Beirut, Beirut, Lebanon.
Address Correspondence and Requests for Single Reprints to: Robert H. Habib, PhD Director, Outcomes Research Unit American University of Beirut-Medical Center PO BOX: 11-0236, Riad El Solh 1107 2020, Beirut, Lebanon Tel:+961-1-374444 ext.5446 E-mail: [email protected] Word Count: 3,495 words
1
ACCEPTED MANUSCRIPT
1 2 3 4 5
Abstract
6
after coronary artery bypass grafting (CABG) is confounded by the frequent concomitant serious
7
complications that co-occur with POAF. We aimed to define the magnitude and time-
8
dependence of the effect of isolated POAF on late survival after uncomplicated CABG in a
9
manner that comprehensively accounts for comorbidity and perioperative confounding factors.
10
Methods: Non-salvage CABG patients with no history of AF, no concomitant aortic/valvular
11
surgery, and no perioperative complications other than POAF were studied (N=6305). Patients
12
were divided to AF (N=1211, 68 yrs, 72% male) and No-AF (N=5094, 63 yrs, 70% male)
13
groups. Propensity matching was done based on 55 patient variables including coronary grafts,
14
completeness of revascularization and transfusion data. The AF-effect was quantified using time-
15
segmented hazard ratios (HRt) by Cox regression.
16
Results: Single (1-to-1), double (1-to-2) and triple (1-to-3) propensity matching of AF to No-AF
17
was achieved in 1196, 993 and 719 cases, respectively. AF showed significantly worse, yet time-
18
varying, 0-18 year survival: (0-1 yrs: HR=1.18 (0.77-1.81), 1-6 yrs: HR=1.37 (1.12-1.67), 6-17
19
yrs: 1.25 (1.05-1.49)).
20
Conclusions: Isolated POAF is associated with a time-varying increase in mortality after CABG
21
surgery. Given these findings and the high incidence of POAF, efforts to reduce POAF should be
22
pursued to potentially improve resource utilization, morbidity and mortality.
RI PT
SC
M AN U
TE D
EP
AC C
23
Objectives: The association of new-onset postoperative atrial fibrillation (POAF) and late death
24 25 26 2
ACCEPTED MANUSCRIPT
1
Ultra mini-abstract
3
This study investigated long term outcomes of new onset atrial fibrillation (POAF) following
4
coronary artery bypass surgery (CABG). Using multiple propensity matching methods, we found
5
that POAF increases long term mortality over an 18 year postoperative follow up period.
RI PT
2
AC C
EP
TE D
M AN U
SC
6
3
ACCEPTED MANUSCRIPT
1 2
Introduction New onset postoperative atrial fibrillation (POAF) is a frequent complication of cardiac (10%-65%) surgery1-6, and it is associated with increased morbidity, in-hospital mortality, and
4
greater resource utilization.1-10 Imbalance of the autonomic nervous system, enhanced
5
inflammation, ischemia and predisposing anatomic substrate has been implicated in PAOF,
6
however, the exact pathophysiology of POAF is poorly understood. The incidence of POAF in
7
coronary bypass grafting (CABG: 15%-40%) is reportedly predicted by patient demographics
8
including older age, lower ejection fraction, hypertension, heart failure, prolonged
9
cardiopulmonary bypass, enlarged left atrium, perioperative medications and enhanced
11
SC
inflammatory and oxidative stress.7-10
M AN U
10
RI PT
3
A number of recent studies have extended the adverse effects of POAF after CABG to include significantly worse late mortality.2, 9, 11-16 If true, this finding is of substantial clinical
13
importance given the high incidence of POAF, yet it is not clear what is the specific mechanism
14
by which an ostensibly limited perioperative arrhythmia, such as POAF, affects long term
15
survival. The available evidence on this POAF - late mortality association is limited by non-
16
uniformity of analyses and findings and/or incomplete accounting for known determinants of late
17
mortality after CABG. No studies, thus far, account for a) extent of arterial grafting and b)
18
completeness of coronary revascularization.17-22 Furthermore, homologous blood product
19
transfusion, which is notably more frequent in patients with higher likelihood of POAF, and has
20
been identified as a predictor of late mortality has also been largely ignored in studies of the
21
effects of POAF.23-25 In addition, POAF is frequently associated with other serious postoperative
22
complications and, consequently the interplay between these non-arrhythmic complications and
23
POAF may confound the independent effects of these conditions on operative and possibly late
AC C
EP
TE D
12
4
ACCEPTED MANUSCRIPT
death.2, 8, 9, 11 Separating the late effects of POAF and concurrent complications is difficult and
2
studies, thus far, have either ignored this 13, 26 or have relied on multivariate Cox regression
3
modeling to account for some but not all such complications.2, 9, 11, 14, 27 Lastly, the potentially
4
time-varying POAF effect on late mortality, its duration and the time period of highest risk, have
5
not been described.
6
RI PT
1
Within this construct, we reasoned that if POAF independently leads to increased late CABG mortality, then it should do so even in 1) the absence of other non-arrhythmic
8
postoperative complications, and 2) after more comprehensive adjustment for other known pre,
9
intra, and post-operative predictors of late survival. Accordingly, we investigated the POAF late
M AN U
SC
7
10
CABG survival association in a manner that avoids the above-mentioned limitations by more
11
comprehensive accounting for patient variables, operative and grafting details, transfusion status
12
and most importantly by excluding patients with other complications.
14
TE D
13
Methods
This investigation is a retrospective analysis of a prospectively collected Society of
16
Thoracic Surgeons (STS) Adult Cardiac Surgery National Database approved by the Mercy Saint
17
Vincent Medical Center (MSVMC, Toledo, OH; 1994-2007) Institutional Review Board. Data
18
collection was performed in compliance with the STS Database definitions and criteria and did
19
not involve additional review of hospital records or interview of patients, thus informed consent
20
requirement was waived.
22
AC C
21
EP
15
Patients. Patients undergoing isolated CABG between 1994 and 2007 were included in
23
the analysis. CABG patients with concurrent coronary or carotid reconstruction were also
24
included. Patients were excluded in case of concomitant valvular, congenital cardiac or aortic 5
ACCEPTED MANUSCRIPT
surgery, documented history of preoperative atrial fibrillation/flutter, or emergency-salvage
2
status. A total of 7,610 patients qualified for inclusion, and these were subsequently grouped as
3
follows: (1) 5,094 patients who did not develop POAF or any other postoperative complication
4
(No-AF; control group); (2) 1,211 patients with isolated POAF without other complications (AF;
5
study group); (3) 832 patients without POAF but who developed other complications (No-AF/C);
6
and (4) 473 patients who developed POAF and other complications (AF/C). To mitigate any
7
potential confounding effects of concurrent non-rhythm related perioperative complications on
8
long term survival, the subcohort of patients without any other perioperative complications aside
9
from POAF was selected as the principal focus of the investigation. Time zero was defined as
M AN U
SC
RI PT
1
time of discharge. In hospital deaths were counted as patients with complications and were
11
therefore excluded from the AF and No-AF groups. POAF was defined, as per the STS Database,
12
as the occurrence of POAF or atrial flutter requiring treatment (i.e. beta blockers, calcium
13
channel blockers, amiodarone, anticoagulation, or cardioversion) throughout hospitalization or at
14
readmission within 30 days postoperatively. Telemetry was used in 100% of patients during
15
hospitalization.
TE D
10
17
EP
16
Surgery. Our surgical techniques have been previously reported 17, 19. Cardiopulmonary bypass (96%) and aortocoronary grafting (97%) were used in a large majority of patients.
19
Evidence-based critical pathways were utilized and periodically updated in compliance with
20
emerging data documented to optimize outcomes, improve care or decrease resource utilization.
21
Treatment of POAF, if it occurred, is described in the On Line Supplement. All efforts were
22
made to discharge patients in a normal sinus rhythm. Patients discharged in rate controlled atrial
AC C
18
6
ACCEPTED MANUSCRIPT
1
fibrillation were routinely discharged on warfarin, unless contraindicated. The database is not
2
granular enough to derive the subset of POAF patients discharged in a non sinus rhythm.
3
Perioperative Medications. Aspirin and β-blockers use was consistent throughout the
RI PT
4
1994 to 2007 study period. In all patients, β-blockers were routinely administered orally until the
6
morning of surgery and in stable patients were reinstituted on postoperative day 1. Accordingly,
7
time of β-blocker withdrawal (if any) was minimized in the present series of patients. An ACE
8
inhibitor was used routinely in patients with normal renal function, including those who were
9
taking one preoperatively, who had a history of congestive heart failure or who had low ejection
M AN U
SC
5
fraction (100 mm Hg) in the
11
early phases of the study period and routinely in the later phase of the study period in compliance
12
with practice guidelines. Aspirin (325 mg via nasogastric tube) was always given to patients on
13
arrival at the Cardiovascular Intensive Care Unit and was decreased to 162 mg PO daily on
14
postoperative day 1. Amiodarone (intravenous or orally), once approved for clinical use, was
15
routinely used after onset of AF, including intra-operatively. Starting July 1999, prophylactic
16
amiodarone was used routinely in all patients (daily 200 mg PO started on postoperative day 1).
17
Prior to the availability of amiodarone, quinidine and/or procainamide in conjunction with
18
calcium channel antagonists were used to treat POAF.
20
EP
AC C
19
TE D
10
Follow-up. All-cause mortality was secured from the cardiothoracic surgery follow-up
21
and verified from scheduled searches of the US Social Security Death Index database
22
(http://ssdi.genealogy.rootsweb.com) last performed in November 2011. Follow-up was between
23
56 (minimum) and 214 (maximum) months.
7
ACCEPTED MANUSCRIPT
Propensity Score Model and Matching. AF and No-AF groups exhibited significant
2
demographic and risk factor differences (Table 1, Table-S1 of Supplement). Such differences
3
confound outcome comparisons in observational comparison groups. To minimize such
4
confounding, we used propensity score matching to derive demographics, risk factor and
5
operation matched sub-cohorts of equal size. The propensity score, or equivalently the
6
probability that a given patient develops POAF, was derived via a non-parsimonious logistic
7
multivariate regression model that considered POAF status (AF or No-AF) as the dependent
8
outcome variable. A total of 55 preoperative risk factors (Tables 1, S1), demographics and
9
operative variables were entered into the model irrespective of significance. A completeness of
10
revascularization index (CRI= #Grafts-VD) derived from the difference between the number of
11
grafts (#Grafts or distal anastomoses in the case of sequential grafts) and coronary vessel disease.
12
Time-of-surgery was entered as a continuous year of series variable [1994=1, 2007 =14] to
13
account for potential variance in reporting POAF or its treatment. Highly redundant variables
14
were avoided. The resulting propensity scores were distinctly different for isolated AF versus
15
No-AF patients [mean±SD: 0.259±0.124 vs. 0.176±0.106 respectively; p 40 kg/m2) Smoker Diabetes Insulin Dependent Renal Failure Hypercholesterolemia Hypertension Chronic Lung Disease Peripheral Vascular Disease Cerebrovascular Disease
Matched CABG patients No-AF AF (n=1196) (n=1196) Mean ± SD Mean ± SD 67.6±9.5 67.5±9.5 2.04±0.25 2.04±0.25 7.9±3.6 7.8±3.5
50±11 79±35 3.2±1 0.54±0.79
EP
Ejection Fraction (EF, %) Perfusion Time (min) Number of Grafts Revascularization Index
AC C
Variable Continuous Age (yrs) Body surface area (BSA, m2) Year of surgeryΩ
Un-matched CABG patients No-AF AF (n=5094) (n=1211) Mean ± SD Mean ± SD p-value 62.5±10.6 67.6±9.6
