Research

Original Investigation

Preoperative β-Blocker Use in Coronary Artery Bypass

Grafting Surgery
National Database Analysis
William Brinkman, MD; Morley A. Herbert, PhD; Sean O’Brien, PhD; Giovanni Filardo, PhD; Syma Prince, RN;
Todd Dewey, MD; Mitchell Magee, MD; William Ryan, MD; Michael Mack, MD

Invited Commentary
IMPORTANCE Use of preoperative β-blockers has been associated with a reduction in page 1328
perioperative mortality for patients undergoing coronary artery bypass grafting (CABG) Supplemental content at
surgery in observational research studies, which led to the adoption of preoperative jamainternalmedicine.com
β-blocker therapy as a national quality standard.

OBJECTIVE To determine whether preoperative β-blocker use within 24 hours of CABG

surgery is associated with reduced perioperative mortality in a contemporary sample of
patients.

DESIGN, SETTING, AND PARTICIPANTS Retrospective analysis of the Society of Thoracic

Surgeons National Adult Cardiac database for 1107 hospitals performing cardiac surgery in the
United States from January 1, 2008, through December 31, 2012. Participants included
506 110 patients 18 years and older undergoing nonemergent CABG surgery who had not
experienced a myocardial infarction in the prior 21 days or any other high-risk presenting
symptom. We used logistic regression and propensity matching with a greedy 5-to-1
digit-matching algorithm to examine the association between β-blocker use and the main
outcomes of interest.

EXPOSURES Preoperative β-blocker use.

MAIN OUTCOMES AND MEASURES Incidence of perioperative mortality, permanent stroke,

prolonged ventilation, any reoperation, renal failure, deep sternal wound infection, and atrial
fibrillation.

RESULTS Among the 506 110 patients undergoing CABG surgery who met the inclusion
criteria, 86.24% received preoperative β-blockers within 24 hours of surgery. In
propensity-matched analyses that included 138 542 patients, we found no significant
difference between patients who did and did not receive preoperative β-blockers in rates of
operative mortality (1.12% vs 1.17%; odds ratio [OR], 0.96 [95% CI, 0.87-1.06]; P = .38),
permanent stroke (0.97% vs 0.98%; OR, 0.99 [95% CI, 0.89-1.10]; P = .81), prolonged
ventilation (7.01% vs 6.86%; OR, 1.02 [95% CI, 0.98-1.07]; P = .26), any reoperation (3.60%
vs 3.69%; OR, 0.97 [95% CI, 0.92-1.03]; P = .35), renal failure (2.33% vs 2.24%; OR, 1.04 Author Affiliations:
[95% CI, 0.97-1.11]; P = .30), and deep sternal wound infection (0.29% vs 0.34%; OR, 0.86 Cardiopulmonary Research Science
and Technology Institute, Dallas,
[95% CI, 0.71-1.04]; P = .12). However, patients who received preoperative β-blockers within Texas (Brinkman, Prince, Dewey,
24 hours of surgery had higher rates of new-onset atrial fibrillation when compared with Magee, Ryan, Mack); Department of
patients who did not (21.50% vs 20.10%; OR, 1.09 [95% CI, 1.06-1.12]; P < .001). Results of Clinical Research, Medical City Dallas
Hospital, Dallas, Texas (Herbert,
logistic regression analyses were broadly consistent.
Dewey, Magee); Duke Clinical
Research Institute, Durham, North
CONCLUSIONS AND RELEVANCE Preoperative β-blocker use among patients undergoing Carolina (O’Brien); Institute for
nonemergent CABG surgery who have not had a recent myocardial infarction was not Health Care Research and
Improvement, Baylor Health Care
associated with improved perioperative outcomes. System, Dallas, Texas (Filardo).
Corresponding Author: William
Brinkman, MD, Cardiopulmonary
Research Science and Technology
Institute, 4716 Alliance Blvd,
JAMA Intern Med. 2014;174(8):1320-1327. doi:10.1001/jamainternmed.2014.2356 Pavilion 2, Ste 310, Plano, TX 75093
Published online June 16, 2014. (willibri@baylorhealth.edu).

1320 jamainternalmedicine.com

Copyright 2014 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015

 Preoperative β-Blocker Use in CABG Surgery
β
-Blockers have been used perioperatively in cardiovas-
cular surgery for more than 40 years.1 Strategies for the
use of β-blockers before cardiac surgery have changed
over time. For many years these therapies were stopped
before surgery owing to concerns about negative inotropic ef-
fects. However, in the late 1990s, multiple retrospective analy-
ses demonstrated benefits of preoperative β-blockade use.2,3
Since 2007, the use of preoperative β-blockers has been used
as a quality standard for patients undergoing coronary artery
bypass grafting (CABG) surgery4,5; hospital and surgeon reim-
bursements may be affected by meeting goals for this
standard.6
Although initial studies (based on CABG data from the
1990s) provided support,7 no randomized prospective analy-
ses of the use of preoperative β-blockers in CABG have been
performed. Retrospective and prospective randomized stud-
ies in the noncardiac surgical literature have failed to show a
benefit with their routine empirical application.8-11 Some in-
vestigators have hypothesized that the benefit of β-blockers
observed in early studies may been driven by those with re-
cent myocardial infarctions (MIs), a cohort known to benefit
from aggressive β-blockade. A meta-analysis of 69 studies by
Wiesbauer et al12(p35) reported that perioperative β-blockers
“had no effect on the frequency of hard end-points like myo-
cardial infarction, mortality, or length of hospitalization” in car-
diac and noncardiac surgery. A recent analysis of Society of
Thoracic Surgeons National Adult Cardiac database (STS-
NCD) data collected from a group of North Texas hospitals dem-
onstrated no benefit associated with empirical preoperative
β-blocker use before CABG in a cohort that excluded these
patients.13 In addition, the principal data used for mortality ben-
efit associated with β-blockade in noncardiac surgery, the
Dutch Echocardiographic Cardiac Risk Evaluation Applying
Stress Echocardiography (DECREASE), has been shown re-
cently to be falsified and was retracted in a highly publicized
episode of academic misconduct.14 The present study as-
sessed whether administration of β-blockers in patients with-
out a recent MI (ie, within 21 days from CABG) was associated
with reduced operative mortality and in general reduced in-
cidence of postoperative adverse outcomes.
Methods
The STS-NCD was established in 1989 for the purpose of out-
come assessment after cardiac surgery. Data are collected quar-
terly from participating hospitals in the United States and
Canada, with strict data quality standards. These data are then
warehoused at the Duke Clinical Research Institute and used
to generate site-level quality assurance and quality improve-
ment feedback reports. Regular audits of the participating cen-
ters are used to measure data accuracy and integrity. To as-
sess the association between preoperative β-blocker use and
postoperative adverse outcomes, we used STS-NCD data.
Study Cohort
The study cohort consisted of patients 18 years and older
who underwent nonemergency isolated CABG surgery at
jamainternalmedicine.com
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015
Original Investigation Research
STS-NCD–participating hospitals from January 1, 2008,
through December 31, 2012. From a starting population
undergoing 756 215 operations, we excluded patients with
prior MI within 21 days (211 715 [28.00%]); patients with
high-risk presenting symptoms, including shock, prior per-
cutaneous coronary intervention within 6 hours, or preop-
erative intra-aortic balloon pump or inotropes (18 894
[2.50%]); patients with a documented contraindication to
β-blocker therapy (18 881 [2.50%]); and patients with miss-
ing data on key variables, including age, sex, and β-blocker
use (615 [0.08%]). The final study cohort included 506 110
patients from 1107 participating centers. The type of analy-
ses presented in this study has been reviewed and approved
by the Duke University Health System institutional review
board under protocol number CR1_Pro00005876. Informed
consent was waived.
Exposure Definition
Preoperative β-blocker use was defined as receiving a dose of
a β-blocker within 24 hours preceding surgery (as per STS-
NCD definition1; STS data definitions for all data fields are found
at http://www.sts.org/sites/default/files/documents/word
/STSAdultCVDataSpecificationsV2_73%20with%20correction
.pdf for version 2.73 and at http://www.sts.org/sites/default
/ f i l e s /d o c u m e n t s / p d f / t r a i n i n g m a n u a l s /a d u l t 2 .6 1
/V-c-AdultCVDataSpecifications2.61.pdf for version 2.61).
According to STS-NCD data definitions, contraindications to
β-blocker therapy were only counted if they were docu-
mented in the medical record by a physician, a nurse practi-
tioner, or a physician assistant.
Outcome Definitions
The following outcomes (defined by the STS-NCD, version
2.73) were considered for this study. Operative mortality
included (1) all deaths occurring during the acute episode of
care in which the operation was performed and (2) those
deaths occurring after discharge from the hospital, but
within 30 days of the procedure. Stroke was defined as any
confirmed neurologic deficit of abrupt onset caused by a
disturbance in the blood supply to the brain that did not
resolve within 24 hours. Prolonged ventilation included any
pulmonary ventilator use for longer than 24 hours. Any
reoperation included reoperation for bleeding, graft occlu-
sion, valvular dysfunction, or other cardiac reasons. Renal
failure included acute or worsening renal function resulting
in (1) an increase of serum creatinine level to at least 3 times
the most recent preoperative level or a serum creatinine
level of at least 4.0 mg/dL with an minimum increase of 0.5
mg/dL relative to the last preoperative value (to convert to
micromoles per liter, multiply by 88.4) and/or (2) a new
postoperative requirement for dialysis. Deep sternal wound
infection included any wound infection within 30 postop-
erative days involving muscle, bone, and/or mediastinum
requiring operative intervention or a positive culture find-
ing requiring antibiotic treatment. Atrial fibrillation
included a new onset of atrial fibrillation/flutter requiring
treatment, excluding recurrence of previously documented
preoperative atrial fibrillation/flutter.
JAMA Internal Medicine August 2014 Volume 174, Number 8 1321
 Research Original Investigation
Table 1. β-Blocker Use Overall and by Year of Operation
2008 2009
No. of patients 111 059 109 178
No. (%) receiving β-blockers 89 012 (80.15) 90 819 (83.18)
Definition of Baseline Covariates
Body surface area (BSA) was calculated by the following Du Bois
formula:
BSA = Weight (in Kilograms)0.425 × Height (in
Centimeters)0.725
Other clinical factors were defined according to STS-NCD data
specifications, using version 2.61 (January 1, 2008, through June
30, 2011) or version 2.73 (July 1, 2011, through December 31,
2012). Atrial fibrillation was defined as preoperative fibrilla-
tion or flutter within 2 weeks of operation using version 2.61
or within 30 days of operation using version 2.73. Other pre-
operative factors had identical definitions across the 2 data ver-
sions or were mapped based on established conventions for
the STS risk models.
Statistical Analysis
Baseline demographic and clinical characteristics were sum-
marized for patients receiving and not receiving β-blockers (β-
blocker and non–β-blocker groups, respectively). Baseline fac-
tors were summarized as percentages for categorical variables
and as mean (SD) for continuous variables. Statistical testing
was not conducted for baseline factors because the very large
sample size causes even small differences to be highly statis-
tically significant. Analysis was performed using commer-
cially available software (SAS, version 9.3; SAS Institute Inc).
We used the following 2 techniques to adjust for selec-
tion bias when comparing outcomes of the β-blocker vs non–
β-blocker groups: multivariable regression modeling and pro-
pensity matching. For the regression-based analyses, the
association between β-blocker use and each clinical end point
was estimated in a logistic regression model with adjustment
for the following patient-level covariates: age, body surface
area, sex, Hispanic or nonwhite race, dyslipidemia, prior CABG,
prior percutaneous coronary intervention, 2 or more prior car-
diovascular operations, hypertension, immunosuppressive
therapy, peripheral vascular disease, unstable angina, left main
coronary artery disease, number of diseased vessels (<2, 2, or
3), cerebrovascular disease, prior cerebrovascular accident, dia-
betes mellitus (types 1 or 2 or none), urgent status, congestive
heart failure (New York Heart Association class IV, classes I-
III, or none), atrial fibrillation, ejection fraction, chronic lung
disease (severe, moderate, mild, or none), dialysis, creati-
nine level, and prior MI. In addition to patient-level covari-
ates, each logistic model also included year of surgery to ac-
count for temporal trends in outcomes and β-blocker use and
a set of fixed-effect hospital-specific intercept variables to con-
trol for potential confounding by center effects.15-18 Missing
data occurred in fewer than 0.5% of records for all model co-
variates except ejection fraction (3.22%), unstable angina
(0.78%), and mitral insufficiency (0.67%). Missing values for
continuous covariates were imputed by stratifying patients by
1322 JAMA Internal Medicine August 2014 Volume 174, Number 8
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015
Preoperative β-Blocker Use in CABG Surgery
2010 2011 2012 All
103 234 93 322 89 317 506 110
89 425 (86.62) 84 403 (90.44) 82 817 (92.72) 436 476 (86.24)
combinations of related risk factors and then imputing stratum-
specific medians. Missing values for categorical covariates were
imputed to the most common category. Alternative sophisti-
cated missing data techniques (ie, multiple imputation) were
not used for this analysis owing to the low rate of missing data
and because multiple imputation has had a negligible effect
on previous analyses of the STS-NCD.
The second method of adjusting for selection bias in-
volved matching patients with similar estimated probability
of receiving β-blockers (ie, their propensity score).19 To esti-
mate the propensity score, a logistic regression model predict-
ing the use of preoperative β-blockers was developed using the
same covariates listed above for the regression-based analy-
ses. With the exception of ejection fraction, handling of miss-
ing data was identical to that for the regression-based analy-
ses. For ejection fraction, the propensity model was augmented
to include a missing data indicator variable. Inclusion of this
variable ensured an equal prevalence of missing data for ejec-
tion fraction for patients with and without β-blockers in the
propensity-matched sample. After estimating propensity
scores, we matched patients using a greedy 5-to-1 digit-
matching algorithm.20 Before analyzing outcomes, we as-
sessed the success of the propensity-matching procedure by
comparing the distribution of patient characteristics in the
matched sample. Finally, odds ratios (ORs) comparing the fre-
quency of each end point for patients receiving vs not receiv-
ing β-blockers were estimated using univariable logistic re-
gression and reported with 95% CIs.
In addition to comparing overall outcomes for patients re-
ceiving vs not receiving β-blockers, we further examined
whether the association between β-blockers and mortality dif-
fered across prespecified subgroups based on age, sex, con-
gestive heart failure, ejection fraction, diabetes mellitus, and
chronic lung disease. For each subgroup, the association with
mortality was estimated in a logistic regression model that in-
cluded a treatment group indicator (0, no β-blockers; 1, β-
blockers), an indicator variable for the subgroup of interest,
and the interaction between the treatment group and sub-
group indicators. Subgroup-specific ORs were estimated and
displayed with 95% CIs, and we used a test of treatment-by-
subgroup interaction to assess whether differences in the es-
timated ORs across subgroups were consistent with chance.
Results
The study cohort included 506 110 patients who underwent non-
emergency isolated CABG from January 1, 2008, through De-
cember 31, 2012, and did not have a prior MI within 21 days of
operation or other high-risk presenting symptoms. The use of
preoperative β-blockade in the STS-NCD changed from 80.15%
in 2008 to 92.72% in 2012 (Table 1). Rates of use varied from 20%
jamainternalmedicine.com
 Preoperative β-Blocker Use in CABG Surgery Original Investigation Research

Table 2. Baseline Demographic and Clinical Characteristics in Overall Cohorta

β-Blocker Use Groups

All Patients No Yes
Variable (n = 506 110) (n = 69 634) (n = 436 476)
Age, mean (SD), y [No. with data] 65.0 (10.3) [506 110] 65.7 (10.1) [69 634] 64.9 (10.3) [436 476]
Male sex 374 526/506 110 (74.00) 52 698/69 634 (75.68) 321 828/436 476 (73.73)
Nonwhite race or Hispanic ethnicity 83 843/504 707 (16.61) 11 329/69 376 (16.33) 72 514/435 331 (16.66)
Operation after 2010 182 639/506 110 (36.09) 15 419/69 634 (22.14) 167 220/436 476 (38.31)
BSA, mean (SD), m2 [No. with data] 2.0 (0.2) [504 328] 2.0 (0.2) [69 261] 2.0 (0.2) [435 067]
Diabetes mellitus 213 301/505 898 (42.16) 27 483/69 620 (39.48) 185 818/436 278 (42.59)
Portion treated with insulin 71 068/505 622 (14.06) 8416/69 578 (12.10) 62 652/436 044 (14.37)
Dyslipidemia 440 197/505 402 (87.10) 57 612/69 508 (82.89) 382 585/435 894 (87.77)
Hypertension 443 990/505 966 (87.75) 57 003/69 623 (81.87) 386 987/436 343 (88.69)
Dialysis 11 605/504 396 (2.30) 1275/69 392 (1.84) 10 330/435 004 (2.37)
Last creatinine level, mean (SD), mg/dL 1.1 (0.4) [493 559] 1.0 (0.4) [68 056] 1.1 (0.4) [425 503]
[No. with data]b
Chronic lung disease, moderate or worse 45 596/505 311 (9.02) 6037/69 530 (8.68) 39 559/435 781 (9.08)
Comorbid disease
Severe chronic lung 16 149/505 311 (3.20) 2010/69 530 (2.89) 14 139/435 781 (3.24)
Peripheral vascular 76 078/505 733 (15.04) 10 175/69 600 (14.62) 65 903/436 133 (15.11)
Cerebrovascular 72 994/505 822 (14.43) 9636/69 611 (13.84) 63 358/436 211 (14.52)
CVA 33 629/505 114 (6.66) 4363/69 499 (6.28) 29 266/435 615 (6.72)
CHF 62 239/504 749 (12.33) 6752/69 473 (9.72) 55 487/435 276 (12.75)
CHF (NYHA class IV) 8707/504 749 (1.73) 814/69 473 (1.17) 7893/435 276 (1.81)
Ejection fraction, mean (SD), % 53.5 (11.7) [489 790] 54.9 (10.9) [66 518] 53.3 (11.8) [423 272]
Prior MI 132 288/505 283 (26.18) 12 919/69 503 (18.59) 119 369/435 780 (27.39)
Unstable angina 244 420/502 178 (48.67) 29 110/68 874 (42.27) 215 310/433 304 (49.69)
No. of diseased vessels
2 108 582/505 548 (21.48) 15 890/69 542 (22.85) 92 692/436 006 (21.26)
3 370 068/505 548 (73.20) 49 526/69 542 (71.22) 320 542/436 006 (73.52)
Left main coronary artery disease 155 723/504 788 (30.85) 21 757/69 409 (31.35) 133 966/435 379 (30.77)
Recent atrial fibrillation 24 741/505 801 (4.89) 2831/69 592 (4.07) 21 910/436 209 (5.02)
Prior procedure
CABG 18 517/505 757 (3.66) 1978/69 598 (2.84) 16 539/436 159 (3.79)
PCI 138 769/505 819 (27.43) 14 106/69 602 (20.27) 124 663/436 217 (28.58)
Valve surgery 1475/505 729 (0.29) 177/69 593 (0.25) 1298/436 136 (0.30)
No. of prior CV operations
≥1 20 655/505 762 (4.08) 2232/69 571 (3.21) 18 423/436 191 (4.22)
≥2 1249/505 762 (0.25) 110/69 571 (0.16) 1139/436 191 (0.26)
Mitral insufficiency, moderate or severe 17 051/502 722 (3.39) 1699/69 278 (2.45) 15 352/433 444 (3.54)
Immunosuppressive therapy 11 241/505 684 (2.22) 1459/69 601 (2.10) 9782/436 083 (2.24)
Status
Elective 276 795/506 110 (54.69) 44 062/69 634 (63.28) 232 733/436 476 (53.32)
Urgent 229 315/506 110 (45.31) 25 572/69 634 (36.72) 203 743/436 476 (46.68)
a
Abbreviations: BSA, body surface area; CABG, coronary artery bypass surgery; Unless otherwise indicated, data are expressed as number (percentage) of
CHF, congestive heart failure; CVA, cerebrovascular accident; MI, myocardial patients.
infarction; NYHA, New York Heart Association; PCI, percutaneous coronary b
Excludes patients undergoing dialysis.
intervention.
SI conversion factor: To convert creatinine to micromoles per liter, multiply by
88.4.

to 100% (median, 87.1%) across all individual reporting units and
from 53% to 100% (median, 87.5%) in the subset of units that
had at least 500 records in the study database.
Clinical characteristics of the study cohort are summa-
rized in Table 2. Patients not receiving β-blockers were less
likely to have a prior MI, prior revascularization, urgent sta-
tus, and operation after 2010. Hypertension, dyslipidemia, dia-
betes mellitus, congestive heart failure, history of unstable an-
gina, and dialysis were more common in patients who received
preoperative β-blockers.

jamainternalmedicine.com JAMA Internal Medicine August 2014 Volume 174, Number 8 1323

Copyright 2014 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015

 Research Original Investigation Preoperative β-Blocker Use in CABG Surgery

Table 3. Number of End Point Events and Covariate-Adjusted ORs in Overall Cohorta

No. (%) of Events by Group

End Point No β-Blocker β-Blocker OR (95% CI)b P Value
Operative mortality 817 (1.17) 5310 (1.22) 0.96 (0.88-1.04) .27
Permanent stroke 682 (0.98) 4441 (1.02) 0.99 (0.91-1.08) .95
Prolonged ventilation 4767 (6.85) 31 305 (7.17) 0.99 (0.96-1.03) .61
Reoperation 2567 (3.69) 15 891 (3.64) 0.96 (0.92-1.01) .11 Abbreviation: OR, odds ratio.
a
Renal failure 1561 (2.24) 9955 (2.28) 1.03 (0.97-1.09) .35 Includes 506 110 patients.
b
Deep sternal infection 237 (0.34) 1319 (0.30) 0.81 (0.70-0.94) .01 Adjusted for 27 patient baseline
factors plus year of operation and
Atrial fibrillation 13 997 (20.10) 93 335 (21.38) 1.11 (1.09-1.14) <.001
hospital characteristic.

Table 4. Number of End Point Events and ORs in Propensity-Matched Samplea

No. (%) of Events by Group

End Point No β-Blocker β-Blocker OR (95% CI) P Value
Operative mortality 813 (1.17) 778 (1.12) 0.96 (0.87-1.06) .38
Permanent stroke 680 (0.98) 671 (0.97) 0.99 (0.89-1.10) .81
Prolonged ventilation 4749 (6.86) 4855 (7.01) 1.02 (0.98-1.07) .26
Reoperation 2554 (3.69) 2489 (3.60) 0.97 (0.92-1.03) .35
Renal failure 1555 (2.24) 1613 (2.33) 1.04 (0.97-1.11) .30 Abbreviation: OR, odds ratio.
a
Deep sternal wound infection 236 (0.34) 203 (0.29) 0.86 (0.71-1.04) .12 Includes 138 542 patients in
propensity-matched samples of
Atrial fibrillation 13 930 (20.10) 14 896 (21.50) 1.09 (1.06-1.12) <.001
69 271 each.

Table 3 summarizes the clinical end points and risk-
adjusted ORs in the nonpropensity-matched sample for pa-
tients receiving vs not receiving preoperative β-blockers. The
observed frequency of operative mortality was similar for both
cohorts (1.22% vs 1.17% for the β-blocker vs non–β-blocker
groups; adjusted OR, 0.96 [95% CI, 0.88-1.04]). Compared with
patients not receiving β-blockers, those receiving β-blockers
had a higher frequency of new-onset postoperative atrial fi-
brillation (21.38% for the β-blocker group vs 20.10% for the non–
β-blocker group; adjusted OR, 1.11 [95% CI, 1.09-1.14]) and a
lower frequency of deep sternal wound infection (0.30% for
the β-blocker vs 0.34% for the non–β-blocker groups; 0.81 [0.70-
0.94]). For other end points, estimated ORs ranged from 0.96
to 1.03 and were not statistically distinguishable from the null
value of 1.00 (for each, P > .10).
After propensity matching, comparable groups of 69 271
patients each were created. As shown in the eTable in the
Supplement, the distribution of measured risk factors was simi-
lar for the β-blocker compared with the non–β-blocker groups.
For each variable in the eTable (Supplement), the difference
in prevalence across the 2 groups was less than 0.5%.
Clinical end points for propensity-matched patients re-
ceiving vs not receiving β-blockers are summarized in Table 4.
The frequency of operative mortality was similar for the 2
groups (1.12% in the β-blocker vs 1.17% in the non–β-blocker
groups; adjusted OR, 0.96 [95% CI, 0.87-1.06]). Compared with
patients not receiving β-blockers, those receiving β-blockers
had a higher frequency of new-onset postoperative atrial fi-
brillation (21.50% in the β-blocker vs 20.10% in the non–β-
blocker groups; adjusted OR, 1.09 [95% CI, 1.06-1.12]). We found
no significant statistical difference in the rate of deep sternal
wound infection (0.29% for the β-blocker group vs 0.34% for
the non–β-blocker group; adjusted OR, 0.86 [95% CI, 0.71-
1.04]). For other end points, estimated ORs ranged from 0.97
to 1.04 and were not statistically distinguishable from the null
value of 1.00 (for each, P > .25).
Table 5 summarizes operative mortality as a function of
β-blocker use for prespecified subgroups in the propensity-
matched cohort. For each subgroup, the 95% CI for the OR over-
lapped unity, and the interaction P value was not significant
(P ≥ .10), indicating that observed differences across sub-
groups were consistent with chance.
Discussion
Summary of Findings
We studied perioperative outcomes in 506 110 patients from 1107
centers who did not have an MI within 21 days of an isolated
CABG operation and who underwent surgery from January 1,
2008, through December 31, 2012. After propensity matching,
comparable groups of 69 271 patients each were created. We per-
formed analyses on the unmatched and propensity-matched
data sets, controlling for the preoperative risk factors. No sta-
tistical mortality benefit was associated with the use of preop-
erative β-blockers. Other end points were equal between groups
except for new-onset postoperative atrial fibrillation, which was
significantly higher in the β-blocker group. On the surface, these
observational study data stand in contrast to the endorsed Na-
tional Quality Forum standards5 that mandate preoperative
β-blocker use before CABG procedures.
The use of preoperative β-blockers before CABG has been
established as a quality indicator by the National Quality
Forum.5 These data are then used to assess the performance
of physicians and hospitals participating in public reporting
via the STS-NCD. The present report builds on a previous analy-

1324 JAMA Internal Medicine August 2014 Volume 174, Number 8 jamainternalmedicine.com

Copyright 2014 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015

 Preoperative β-Blocker Use in CABG Surgery Original Investigation Research

Table 5. Mortality ORs for Prespecified Subgroups in Propensity-Matched Samplea

Mortality by Group, No. (%)

No β-Blocker β-Blocker P Value for
Subgroup (n = 69 271) (n = 69 271) OR (95% CI) Interaction
Male sex 533/52 412 (1.02) 505/52 379 (0.96) 0.95 (0.84-1.07)
.81
Female sex 280/16 859 (1.66) 273/16 892 (1.62) 0.97 (0.82-1.15)
Age, y
<65 207/30 057 (0.69) 182/30 209 (0.60) 0.87 (0.72-1.07)
.30
≥65 606/39 214 (1.55) 596/39 062 (1.53) 0.99 (0.88-1.11)
CHF
Yes 176/6829 (2.58) 159/6824 (2.33) 0.90 (0.73-1.12) .55
No 637/62 391 (1.02) 619/62 372 (0.99) 0.97 (0.87-1.09)
Ejection fraction, %
≥30 714/64 467 (1.11) 696/64 492 (1.081) 0.97 (0.88-1.08)
.10
<30 99/4804 (2.06) 82/4779 (1.72) 0.83 (0.62-1.11)
Diabetes mellitus
Yes 360/27 376 (1.32) 364/27 250 (1.34) 1.02 (0.88-1.18)
.28
No 453/41 881 (1.08) 414/41 971 (0.99) 0.91 (0.80-1.04)
Chronic lung disease
Abbreviations: CHF, congestive heart
Yes 278/14 419 (1.93) 272/14 472 (1.88) 0.97 (0.82-1.15)
.76 failure; OR, odds ratio.
No 533/54 749 (0.97) 502/54 647 (0.92) 0.94 (0.83-1.07) a
Includes 138 542 patients.

sis of β-blocker administration before CABG conducted using
STS-NCD data from a large institution.13 In that analysis of
12 855 patients, investigators were unable to show any ben-
efit with the empirical use of preoperative β-blockers. A prior
analysis 7 of STS-NCD data in 2002 showed contrasting
results.
In the present study we analyzed more recent national data
(2008-2012) from the STS-NCD for more than 500 000 pa-
tients undergoing isolated CABG surgery. In 2002 Ferguson et
al7 reported rates of β-blocker use before coronary bypass pro-
cedures at 50% to 60% for the study period ending in 1999.
Since then, the emphasis on meeting this quality standard has
led to an increase in β-blocker use, with the STS-NCD report-
ing a rate of 86.24% for the use of β-blockers before CABG in
the 2008-2012 data.
Because the use of β-blockers is well established after acute
MI,21 we decided to remove that patient population from our
analysis. Our motivating question was “Do physicians have to
give β-blockers to their patients just because they are sched-
uled for an elective CABG?”
Empirical Preoperative β-Blocker Use in Noncardiac Surgery
In 2006, the results of the Metoprolol After Vascular Surgery
(MaVS) trial, a double-blinded placebo-controlled study, were
published.10 Hypotension and bradycardia were more com-
mon in the treatment group. No difference in cardiac events
(ie, cardiac death, nonfatal MI, unstable angina, congestive
heart failure, or arrhythmia requiring treatment) was demon-
strated. The MaVS trial showed no benefit of perioperative
β-blocker use; of greater concern, it showed potential harm.
Other studies have mirrored this finding of the MaVS trial.11,12
In 2008 the Perioperative Ischemic Evaluation (POISE) study
published their findings.9 The POISE study was a prospective
randomized trial of 8351 patients with known coronary dis-
ease or at risk for coronary disease scheduled to undergo non-
cardiac surgery. Patients were randomized to extended-
release metoprolol succinate or placebo. Analysis of the POISE
data revealed that for every 1200 patients treated, metoprolol
would prevent 15 MIs at a cost of 8 excess deaths and 5 dis-
abling strokes. They suggested that the therapy may benefit
only small selective groups of patients.
A recent meta-analysis22 reviewed 22 trials involving 2437
randomized patients undergoing noncardiac surgery. The au-
thors reported that perioperative use of β-blockers was not as-
sociated with statistically significant benefits on cardiovas-
cular outcomes. Further, they reported an increased risk for
bradycardia and hypotension with empirical β-blockade use.
In their view, “the evidence for perioperative β-blockade was
‘insufficient and inconclusive,’ especially in those at low to
moderate risk of adverse cardiovascular outcomes.”22(p320)
Of greater concern, data used to support previous guide-
lines for perioperative β-blockade use appear to have been at
least partially corrupt. This issue has come to light with the dis-
crediting of the DECREASE family of studies.14 In light of this
problem, Bouri et al11 in 2013 performed a meta-analysis of ran-
domized clinical trials of initiation of β-blocker therapy be-
fore noncardiac surgery that excluded the DECREASE data.
Their conclusions stated that “The well-conducted trials in-
dicate a statistically significant 27% increase in mortality from
the initiation of perioperative β-blockade.”11(p456)
Empirical Preoperative β-Blocker Use in Cardiac Surgery
In 2002, Ferguson et al7 reported an observational study as-
sessing β-blocker use and outcomes among patients undergo-
ing isolated CABG from 1996 through 1999. More than 600 000
patients underwent analysis using STS-NCD data. Preopera-
tive β-blocker use was associated with “slightly lower mortal-
ity after adjusting for patient risk and center effects using both
risk adjustment (OR, 0.94; 95% CI, 0.91-0.97) and treatment pro-
pensity matching (OR, 0.97; 95% CI, 0.93-1.00).”7(p2221) How-

jamainternalmedicine.com JAMA Internal Medicine August 2014 Volume 174, Number 8 1325

Copyright 2014 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015

 Research Original Investigation Preoperative β-Blocker Use in CABG Surgery

ever, in patients with left ventricular ejection fractions of less
than 30%, preoperative β-blockers were associated with a trend
toward an increased mortality rate. The authors stated in their
conclusions: “Although these results are quite promising, we
believe that, ideally, they should be confirmed in a large, ran-
domized clinical trial of preoperative β-blocker use. Such a trial
could also further optimize therapy and better determine the
mechanisms underlying this effect.”7(p2227)
In 2007, Wiesbauer et al12 published a meta-analysis of peri-
operative β-blockade use. This analysis included randomized
clinical trials comparing perioperative β-blockers with pla-
cebo or the standard care in cardiac and noncardiac surgery.
The authors concluded: ”In our review, we found that β-block-
ers.…had no effect on the frequency of hard end-points like
perioperative myocardial infarction, operative mortality, or
length of hospitalization.”12(p35)
In the present study, preoperative β-blocker use in CABG
patients was not associated with lower operative mortality in
the whole study population (1.22% in β-blocker group vs 1.17%
in non–β-blocker group; adjusted OR, 0.96 [95% CI, 0.88-
1.04]) or in a propensity-matched group of patients who have
not had a recent MI (<21 days) (1.17% in the non–β-blocker group
vs 1.12% in the β-blocker group; adjusted OR, 0.96 [95% CI, 0.87-
1.06]). Subgroup analysis of the OR for mortality with preop-
erative β-blocker use showed no benefit for patients in rela-
tion to sex, age, presence of diabetes mellitus, congestive heart
failure, low ejection fraction (<30%), or chronic lung disease.
We were unable to demonstrate any benefit in regard to
the incidence of postoperative atrial fibrillation. In fact, we
noted a significant increase in the incidence of postoperative
atrial fibrillation in the group that had received a β-blocker. This
finding seems counterintuitive and is in contrast to guide-
lines for the treatment of atrial fibrillation.23 The guidelines
advocate preoperative β-blocker use as a class I (level of evi-
dence, A) recommendation. However, a closer analysis of the
studies used to generate this recommendation shows that the
timing of initiation of β-blocker therapy varied widely.22 Es-
sentially, the data used to generate these guidelines were a mix
of preoperative, intraoperative, and postoperative adminis-
tration of β-blockers. A possible explanation for this unex-
pected finding in our data could be the incidence of β-blocker
therapy withdrawal in the non–β-blocker cohort. This effect
has been demonstrated previously.24-26 Because the STS-
NCD only collects data on preoperative β-blocker use (within
24 hours of the incision) as a yes/no field, we are unable to de-
termine whether the timing, the dose given, or other unre-
corded covariates may contribute to this observation. Al-
though we adjusted for recent preoperative atrial fibrillation
(version 2.61 records occurrence within 14 days and version 2.73
records it within 30 days before surgery) as a covariate, we were
unable to adjust for chronic atrial fibrillation, and the β-blocker
group might have contained a higher prevalence of prior atrial
fibrillation occurring remotely. Hypotension at the end of sur-
gery due to β-blocker use may also play a role. However, one
must keep in mind the preponderance in the medical litera-
ture of support for the assertion that perioperative β-block-
ade is effective for the prevention of arrhythmias after car-
diac surgery.27,28
Limitations
The actual β-blocker type, dose, and physiologic response are
not available in the STS-NCD. The present findings could be
related to inadequate dosing or drug type. The STS-NCD is lim-
ited to 30-day perioperative outcomes. Any long-term ben-
efit related to β-blockade would not be demonstrated.
Using the STS-NCD allowed analysis of almost every CABG
case in the United States during the sample period. Despite the
use of preoperative β-blockers being a National Quality Fo-
rum quality indicator, we still see only 86.24% of patients re-
ceiving the drug. The nonadherence comes from surgeons who
believe it is wrong to force the medication on all patients. In
some cases of off-pump revascularization, the surgeons be-
lieve that the risks in manipulation of the heart are increased
in the presence of β-blockade. The data form does not record
reasons for refusal (just verifiable medical contraindica-
tions).
Conclusions
In this large cohort study of patients with isolated CABG who
have not had an MI within 21 days, the administration of
β-blockers before CABG was not associated with improved out-
comes in the study group or in equivalent propensity-
matched groups. These data are consistent with recent litera-
ture regarding the empirical perioperative use of β-blockers in
noncardiac surgery that demonstrates no mortality benefit as-
sociated with their use.
β-Blockers are an important and effective tool in the care
of patients undergoing cardiac surgery in specific clinical sce-
narios. However, the empirical use of β-blockers as recom-
mended by the National Quality Forum (without physiologic
goals, ie, adequate clinical drug levels) in all patients before
CABG may not improve outcomes. A prospective randomized
trial with careful attention to adequate dosing and specific drug
type may help to answer this question.

ARTICLE INFORMATION
Accepted for Publication: April 10, 2014.
Published Online: June 16, 2014.
doi:10.1001/jamainternmed.2014.2356.
Author Contributions: Drs Herbert and O’Brien
had full access to all the data in the study and take
responsibility for the integrity of the data and the
accuracy of the data analysis.
Study concept and design: Herbert, Filardo, Prince,
Dewey, Magee, Ryan, Mack.
Acquisition, analysis, or interpretation of data:
Brinkman, Herbert, O’Brien, Filardo, Prince, Dewey,
Magee, Mack.
Drafting of the manuscript: Herbert, O’Brien,
Filardo, Prince, Dewey, Magee.
Critical revision of the manuscript for important
intellectual content: Brinkman, Herbert, Filardo,
Magee, Ryan, Mack.
Statistical analysis: Herbert, O’Brien, Filardo.
Administrative, technical, or material support:
Brinkman, Prince, Magee.
Study supervision: Herbert, Dewey, Magee, Ryan,
Mack.
Conflict of Interest Disclosures: None reported.
REFERENCES
1. Viljoen JF, Estafanous FG, Kellner GA.
Propranolol and cardiac surgery. J Thorac
Cardiovasc Surg. 1972;64(5):826-830.

1326 JAMA Internal Medicine August 2014 Volume 174, Number 8 jamainternalmedicine.com

Copyright 2014 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015

 Preoperative β-Blocker Use in CABG Surgery
2. Zaugg M, Tagliente T, Lucchinetti E, et al.
Beneficial effects from β-adrenergic blockade in
elderly patients undergoing noncardiac surgery.
Anesthesiology. 1999;91(6):1674-1686.
3. ten Broecke PW, De Hert SG, Mertens E,
Adriaensen HF. Effect of preoperative β-blockade
on perioperative mortality in coronary surgery. Br J
Anaesth. 2003;90(1):27-31.
4. O’Brien SM, Shahian DM, DeLong ER, et al.
Quality measurement in adult cardiac surgery, II:
statistical considerations in composite measure
scoring and provider rating. Ann Thorac Surg. 2007;
83(4)(suppl):S13-S26.
5. Society of Thoracic Surgeons. Quality
performance measures. http://www.sts.org/quality
-research-patient-safety/quality/quality
-performance-measures. Accessed May 8, 2014.
6. National Quality Forum. Quality positioning
system. http://www.qualityforum.org/QPS.
Accessed May 8, 2014.
7. Ferguson TB Jr, Coombs LP, Peterson ED;
Society of Thoracic Surgeons National Adult Cardiac
Surgery Database. Preoperative β-blocker use and
mortality and morbidity following CABG surgery in
North America. JAMA. 2002;287(17):2221-2227.
8. Chopra V, Plaisance B, Cavusoglu E, Flanders SA,
Eagle KA. Perioperative β-blockers for major
noncardiac surgery: Primum Non Nocere. Am J Med.
2009;122(3):222-229.
9. Devereaux PJ, Yang H, Yusuf S, et al; POISE
Study Group. Effects of extended-release
metoprolol succinate in patients undergoing
non-cardiac surgery (POISE trial): a randomised
controlled trial. Lancet. 2008;371(9627):1839-1847.
10. Yang H, Raymer K, Butler R, Parlow J, Roberts
R. The effects of perioperative β-blockade: results
of the Metoprolol After Vascular Surgery (MaVS)
study, a randomized controlled trial. Am Heart J.
2006;152(5):983-990.
11. Bouri S, Shun-Shin MJ, Cole GD, Mayet J, Francis
DP. Meta-analysis of secure randomised controlled
trials of β-blockade to prevent perioperative death
in non-cardiac surgery. Heart. 2014;100(6):456-464.
jamainternalmedicine.com
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a NCSU Hunt Library User on 05/15/2015
12. Wiesbauer F, Schlager O, Domanovits H, et al.
Perioperative β-blockers for preventing
surgery-related mortality and morbidity:
a systematic review and meta-analysis. Anesth Analg.
2007;104(1):27-41.
13. Brinkman WT, Herbert MA, Prince SL, et al.
Preoperative β-blocker usage: is it really worthy of
being a quality indicator? Ann Thorac Surg. 2011;92
(3):788-796.
14. Woodhead M. Cardiology update: research
fraud masks deaths from perioperative beta
blockers. http://www.cardiologyupdate.com.au
/latest-news/research-fraud-concealed-deaths
-from-perioperative. Accessed May 8, 2014.
15. Griswold ME, Localio AR, Mulrow C. Propensity
score adjustment with multilevel data: setting your
sites on decreasing selection bias. Ann Intern Med.
2010;152(6):393-395.
16. Localio AR, Berlin JA, Ten Have TR, Kimmel SE.
Adjustments for center in multicenter studies: an
overview. Ann Intern Med. 2001;135(2):112-123.
17. Berlin JA, Kimmel SE, Ten Have TR, Sammel
MD. An empirical comparison of several clustered
data approaches under confounding due to cluster
effects in the analysis of complications of coronary
angioplasty. Biometrics. 1999;55(2):470-476.
18. Neuhaus JM, Kalbfleisch JD. Between- and
within-cluster covariate effects in the analysis of
clustered data. Biometrics. 1998;54(2):638-645.
19. Li F, Zaslavsky AM, Landrum MB. Propensity
score weighting with multilevel data. Stat Med.
2013;32(19):3373-3387.
20. D'Agostino RB Jr, Rubin DB. Estimating and
using propensity scores with partially missing data.
J Am Stat Assoc. 2000;95(451):749-759.
21. Chen ZM, Pan HC, Chen YP, et al; COMMIT
(Clopidogrel and Metoprolol in Myocardial
Infarction Trial) Collaborative Group. Early
intravenous then oral metoprolol in 45 852 patients
with acute myocardial infarction: randomised
placebo-controlled trial. Lancet. 2005;366(9497):
1622-1632.
JAMA Internal Medicine August 2014 Volume 174, Number 8
Original Investigation Research
22. Devereaux PJ, Beattie WS, Choi PT, et al. How
strong is the evidence for the use of perioperative β
blockers in non-cardiac surgery? systematic review
and meta-analysis of randomised controlled trials.
BMJ. 2005;331(7512):313-321.
23. Fuster V, Rydén LE, Cannom DS, et al; American
College of Cardiology/American Heart Association
Task Force on Practice Guidelines; European
Society of Cardiology Committee for Practice
Guidelines; European Heart Rhythm Association;
Heart Rhythm Society. ACC/AHA/ESC 2006
guidelines 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 European Society of
Cardiology Committee for Practice Guidelines
(writing committee to revise the 2001 guidelines
for the management of patients with atrial
fibrillation): developed in collaboration with the
European Heart Rhythm Association and the Heart
Rhythm Society. Circulation. 2006;114(7):e257-e354.
doi:10.1161/CIRCULATIONAHA.106.177292.
24. Crystal E, Connolly SJ, Sleik K, Ginger TJ, Yusuf
S. Interventions on prevention of postoperative
atrial fibrillation in patients undergoing heart
surgery: a meta-analysis. Circulation. 2002;106(1):
75-80.
25. Ali IM, Sanalla AA, Clark V. β-Blocker effects on
postoperative atrial fibrillation. Eur J Cardiothorac
Surg. 1997;11(6):1154-1157.
26. Budeus M, Feindt P, Gams E, et al. β-Blocker
prophylaxis for atrial fibrillation after coronary
artery bypass grafting in patients with
sympathovagal imbalance. Ann Thorac Surg. 2007;
84(1):61-66.
27. Burgess DC, Kilborn MJ, Keech AC.
Interventions for prevention of post-operative atrial
fibrillation and its complications after cardiac
surgery: a meta-analysis. Eur Heart J. 2006;27(23):
2846-2857.
28. Hammon JW Jr, Wood AJ, Prager RL, Wood M,
Muirhead J, Bender HW Jr. Perioperative β blockade
with propranolol: reduction in myocardial oxygen
demands and incidence of atrial and ventricular
arrhythmias. Ann Thorac Surg. 1984;38(4):363-367.
1327