Puskas JD, Williams WH, Mahoney EM, Huber PR, Block PC, Duke PG, Staples JR, Glas KE, Marshall JJ, Leimbach ME, McCall SA, Petersen RJ, Bailey DE, Weintraub WS, Guyton RA. Off-Pump vs Conventional Coronary Artery Bypass Grafting: Early and 1-Year Graft Patency, Cost, and Quality-of-Life OutcomesA Randomized Trial. JAMA. 2004;291(15):1841-1849. doi:10.1001/jama.291.15.1841
Author Affiliations: Divisions of Cardiothoracic Surgery (Drs Puskas, Williams, and Guyton and Mss McCall, Petersen, and Bailey), Cardiology (Drs Huber, Block, Marshall, and Leimbach), and Cardiac Anesthesiology (Drs Duke, Staples, and Glas), Emory University School of Medicine and Emory Center for Outcomes Research (Drs Williams and Weintraub), Atlanta, Ga; and the New England Research Institutes, Watertown, Mass (Dr Mahoney).
Context Previous trials of off-pump coronary artery bypass (OPCAB) have enrolled
selected patients and have not rigorously evaluated long-term graft patency.
A preliminary report showed OPCAB achieved improved inhospital outcomes, similar
completeness of revascularization, and shorter lengths of stay compared with
conventional coronary artery bypass grafting (CABG).
Objective To assess graft patency, clinical and quality-of-life outcomes, and
cost among patients while in the hospital and at 1-year follow-up.
Design, Setting, and Patients Randomized controlled trial of patients unselected for coronary anatomy,
ventricular function, or comorbidities between March 10, 2000, and August
20, 2001, at a US academic center. A total of 200 patients were enrolled;
3 patients were withdrawn after randomization for mitral valve repair or replacement.
Follow-up was complete for 197 patients at 30 days; 185 at 1 year.
Interventions One surgical session consisting of elective OPCAB or CABG with cardiopulmonary
bypass.The surgeon had extensive experience performing off-pump surgery; patients
were subsequently managed by blinded protocols.
Main Outcome Measures Coronary angiography documented graft patency prior to hospital discharge
and at 1 year; health-related quality of life; and cost of the index and subsequent
Results Graft patency was similar for OPCAB and conventional CABG with cardiopulmonary
bypass at 30 days (absolute difference, 1.3%; 95% confidence interval [CI],
−0.66% to 3.31%; P = .19) and at 1 year (absolute
difference, −2.2%; 95% CI, −6.1% to 1.7%; P = .27). Rates of death, stroke, myocardial infarction, angina, and
reintervention were similar at 30 days and 1 year. There were no significant
differences in health-related quality of life. Mean total hospitalization
cost per patient at hospital discharge was $2272 (95% CI, $755-$3732) less
for OPCAB (P = .002) and $1955 (95% CI, −$766
to $4727) less at 1 year (P = .08).
Conclusions In this randomized single-surgeon trial among unselected patients with
angiographic follow-up, OPCAB achieved similar graft patency in the hospital
and at 1 year. Cardiac outcomes and health-related quality of life at 30 days
and 1 year were similar and patients incurred a lower cost. OPCAB may provide
complete revascularization that is durable and cost-effective.
Motivated by morbidity attributable to cardiopulmonary bypass,1- 3 US surgeons performed
approximately 21% of coronary artery bypass operations off-pump in 2002.4 Nonetheless, concern remains about the technical difficulty
of off-pump coronary artery bypass (OPCAB), including the possibility of imprecise
anastomoses and incomplete revascularization compromising patient outcomes.5- 9 Two
prospective, randomized studies10,11 and
all but a few12 retrospective comparisons have
reported significantly fewer grafts in OPCAB. Retrospective studies among
selected patients showing significant benefits of OPCAB over conventional
coronary artery bypass grafting (CABG) with cardiopulmonary bypass for mortality,13,14 morbidity,12- 15 length
of stay, and cost15 have been criticized for
potential bias in patient selection and management.5- 8 Previous
randomized studies among selected low-risk patients11,16- 18 have
reported similar results, but cannot be extrapolated to the general population
of CABG patients. The potential clinical and economic advantages of OPCAB
in unselected patients and the graft patencies that may be achievable are
The Surgical Management of Arterial Revascularization Therapies (SMART)
trial was designed to compare graft patency, clinical outcomes, health-related
quality of life, and costs in unselected patients referred for elective, isolated
CABG surgery and randomized to OPCAB or CABG with cardiopulmonary bypass.
In a preliminary report, patients receiving OPCAB achieved improved inhospital
outcomes, similar completeness of revascularization, and shorter lengths of
stay compared with patients receiving CABG with cardiopulmonary bypass.19 Results for the primary end point, graft patency
prior to hospital discharge and at 1 year, are now reported. Secondary end
points include clinical and health-related quality-of-life outcomes and costs
during 1-year follow-up.
In accordance with the Declaration of Helsinki and with the institutional
review board approval granted by Emory University, 200 patients provided written
informed consent and were enrolled between March 20, 2000, and August 10,
2001. During this period, 465 patients were referred to a single surgeon (J.D.P.)
for isolated primary, elective CABG. An attempt was made to enroll 1 patient
each operative day to facilitate accurate and complete data acquisition.
Consecutive patients referred each day were screened for eligibility
and asked to volunteer until 1 patient had agreed to participate for the following
day. Thus, 297 nonconsecutive patients were asked to participate, representing
64% of all referrals. Patients were not excluded for any pattern of coronary
artery disease, ventricular dysfunction, or any other comorbidity. Only patients
in cardiogenic shock requiring emergency surgery or preoperative intra-aortic
balloon pump (inserted at cardiologists' discretion) were excluded for cardiac
reasons. Ninety-seven refused to participate. The 200 elective patients randomized
comprised 67% of those screened and 43% of all primary elective coronary referrals
during the enrollment period (Figure 1).
All surgery was performed by a single experienced surgeon (J.D.P.) and
all patient management was conducted by a single team. Patients, their families,
referring cardiologists, and nonoperative clinicians were blinded to treatment
strategy for 1 year. Randomization occurred after documentation by the surgeon
of the intended optimal revascularization. Patients were randomly assigned
by means of a computer-generated random number table and were stratified by
sex and diabetic status. Random assignment was performed separately within
each stratum with randomly permuted block sizes of 4 and 6. Three patients
(1 in the CABG with cardiopulmonary bypass group and 2 in the OPCAB group)
were found after randomization to require mitral valve repair or replacement
and were withdrawn.20 Therefore, the study
population included 197 patients.
Patient management was governed by unbiased and criteria-driven printed
protocols that were applied similarly to both groups.19
Incisions, closure, blood conservation, and suture techniques were similar
in all patients. OPCAB was performed with the Medtronic Octopus II or III
stabilizing device (Medtronic, Minneapolis, Minn) for coronary stabilization
and deep pericardial traction sutures for cardiac displacement.21 Technical
details of surgery for both groups in this trial have been described.19 All patients received aspirin daily.
Graft angiography was planned prior to hospital discharge and at 1-year
follow-up. Only patients with renal insufficiency or severe aortic atherosclerosis
were excluded from postoperative angiography for clinical reasons. Three cardiologists,
who were blinded to group assignment, simultaneously reviewed angiograms.
Each graft was viewed in at least 2 orthogonal planes and scored on the worst
appearance of the proximal anastomosis (if any), body of the conduit and distal
anastomosis—generating a FitzGibbon score22 by
which comparisons were made between groups. A Thrombolysis in Myocardial Infarction
flow score23 was also recorded for each graft.
A score of 0 indicates no flow or no perfusion; 1, slow penetration without
perfusion; 2, partial flow or partial perfusion (>1 but <3); 3, complete
and brisk flow or complete perfusion.21 Left
ventriculograms (right anterior oblique projection) from both the preoperative
and 1-year follow-up arteriograms were evaluated by a single blinded cardiologist.
Left ventricular ejection fraction was calculated by quantitative analysis
using the GE Gemnet software (Fairfield, Conn).
All 192 eligible surviving patients completed a 30-day inperson or telephone
interview assessing adverse events, complications, readmissions, and reinterventions.
Of 189 (95.9%) patients surviving 365 days, 178 (94.2%) were similarly interviewed
at 1-year follow-up. Health-related quality of life was measured using the
EuroQol 624 and 36-item Medical Outcomes Short-Form
Health Survey (SF-36)25 self-administered questionnaires,
which were completed preoperatively and postoperatively at 4 to 6 weeks, 6
months, and 1 year.
Hospital charges for the index hospitalization were derived from the
Medicare formulation of the hospital bill, and costs were calculated using
departmental cost-to-charge ratios. Professional costs for the index hospitalization
were obtained from Current Procedural Terminology codes, which were converted
to relative value units using the resource-based relative value scale.26 Current Procedural Terminology relative value units
were summed and converted to dollars using the Medicare conversion factor.
For all rehospitalizations, which tended to occur at other hospitals, diagnosis-related
groups were assigned and hospital costs were estimated using average Medicare
reimbursement rates. Physician costs for rehospitalizations were estimated
as a percentage of hospital costs according to diagnosis-related group and
corresponding Medicare physician cost to hospital cost percentages.27
This trial was designed with adequate power (80%; α level of .05)
to detect an absolute difference in patency rates between groups in either
direction (as small as 5%). All data analysis was according to the intent-to-treat
principle. Discrete data are presented as numbers and percentages; continuous
data are presented as mean (SD). Dichotomous morbidity and mortality outcomes
were analyzed using the Fisher exact test. Analysis of covariance was used
to compare postoperative laboratory values and health-related quality-of-life
scores between treatment groups after adjusting for preoperative levels. Differences
between groups in health-related quality of life were not considered significant
at follow-up unless P<.01 (to adjust for the multiple
time points of assessment). Dichotomous patency outcomes were analyzed using
the Fisher exact test. Overall dichotomous patency rates were analyzed using
generalized estimating equations28 to account
for the potential correlation between patency outcomes within patients. Ordered
categorical outcomes were compared between groups using the Cochran-Mantel-Haenszel χ2 test. Cost data are presented as mean (SD). Confidence intervals and
associated P values for the difference in average
costs between treatment groups were obtained using bootstrap resampling.29 S-PLUS (version 6, Insightful Corp, Seattle, Wash)
and SAS (version 8.02, SAS Institute Inc, Cary, NC) statistical software were
used for statistical analyses.
Baseline characteristics (Table 1) were similar between groups for all measures except previous stroke
(more common in CABG with cardiopulmonary bypass) and Canadian Cardiovascular
Society angina classifications III and IV (more common in OPCAB).
Comparison of risk factors from Table
1 for patients enrolled in the SMART trial compared with those patients
meeting similar inclusion and exclusion criteria and contemporaneously referred
to the study surgeon (J.D.P.) revealed a significant difference only in the
proportion of women—45 (22.8%) of 197 trial patients compared with 90
(33.0%) of 268 nontrial patients, which is a 10.2% absolute negative difference
(95% confidence interval [CI], 2.0%-18.5%; P = .02).
Additional comparison of risk factors from Table 1 for patients referred to the study surgeon and for patients
meeting similar inclusion and exclusion criteria and referred to the 8 other
Emory cardiothoracic surgeons performing CABG with cardiopulmonary bypass
revealed few significant differences. The mean (SD) age of the study surgeon's
patients was 61.3 (10.7) years, which is 1.7 years younger (95% CI, 0.53-2.8
years; P = .004) than the other surgeons' patients
(63.9 [10.8] years). Chronic obstructive pulmonary disease was present in
180 (38.7%) of the study surgeon's 465 patients compared with 334 (17.4%)
of the other surgeons' 1921 patients, which is a positive difference of 21.3%
(95% CI, 17.5%-25.8%; P<.001). Finally, hypertension
was present in 325 (69.9%) of the study surgeon's 465 patients compared with
1468 (76.4%) of the other surgeons' 1921 patients, which is a negative difference
of 6.5% (95% CI, 1.2%-10.1%; P = .01). No other significant
differences between the study surgeon's patients and the patients of the 8
other Emory surgeons were identified.
Three patients assigned to CABG with cardiopulmonary bypass were reassigned
to OPCAB due to severe aortic atherosclerosis; 1 patient assigned to OPCAB
was reassigned to CABG with cardiopulmonary bypass to graft a deeply intramyocardial
coronary artery. Complications within 30 days and 1 year of surgery were similar
between groups (Table 2). Twelve
patients (8 in the OPCAB group and 4 in the CABG with cardiopulmonary bypass
group) were lost to follow-up or withdrew from the study by 1 year. A search
of the US Social Security Death Index30 revealed
that none of these 12 patients had recorded deaths.
Of 197 enrollees, 184 (93.4%) patients had coronary angiography prior
to hospital discharge. Angiographic FitzGibbon scores were similar between
groups for 622 grafts scored (Table 3).
Overall patency rates (FitzGibbon A plus B) were 99.0% for OPCAB and 97.7%
for CABG with cardiopulmonary bypass (P = .22 [Fisher
exact]; P = .32 [generalized estimating equations];
absolute difference, 1.3% [95% CI, −0.66% to 3.31%]; P = .19). Early patency was similar between groups among all arterial
conduits, all venous conduits, and among grafts to each region of the heart.
Thrombolysis in Myocardial Infarction flows were also similar between groups
Ten (1.6%) of 622 grafts were occluded at cardiac catheterization prior
to hospital discharge. Of these, 3 were in the OPCAB group and 7 were in the
CABG with cardiopulmonary bypass group. Nine of the 10 were saphenous vein
conduits to non–left anterior descending coronary targets. None of these
occluded grafts was associated with recurrent angina or Q-wave myocardial
Of the 189 patients who were alive at 1 year, 153 (81%) had coronary
angiography a mean (SD) of 384.5 (32) days after surgery (OPCAB group: 385.8
; CABG with cardiopulmonary bypass group: 383.3 ). Seven were deemed
ineligible due to renal insufficiency (creatinine level >2.5 mg/dL [221 µmol/L]),
other medical problems, or social circumstances (3 in OPCAB group and 4 in
CABG with cardiopulmonary bypass group) and 17 patients refused follow-up
angiography (7 in OPCAB group and 10 in CABG with cardiopulmonary bypass group).
Angiographic FitzGibbon scores were similar between groups for 511 grafts
scored (Table 3; absolute difference
−2.2%; 95% CI, −6.1% to 1.7%; P = .27).
Overall, 93.6% of grafts were patent among OPCAB patients compared with 95.8%
of the grafts among CABG with cardiopulmonary bypass patients (P = .33 [Fisher exact]; P = .44 [generalized
estimating equation]). Patency was similar between groups at 1 year among
all arterial conduits, all venous conduits, and among grafts to each region
of the heart. Thrombolysis in Myocardial Infarction flows were similar between
groups at 1 year (Table 4).
Twenty-seven (5.3%) of 511 grafts studied were occluded at 1-year follow-up.
Of these, 16 were in OPCAB patients and 11 were in CABG with cardiopulmonary
bypass patients. Of these 27 occluded grafts, 19 were saphenous vein conduits
to non–left anterior descending coronary targets, while 4 were radial
artery grafts and 4 were right or left internal mammary artery grafts.
There were no significant differences between groups in the incidence
of death, myocardial infarction, stroke, recurrent angina, readmission for
cardiac or noncardiac events, or percutaneous reintervention during hospitalization,
at 30 days, or at 1-year follow-up. As of January 2004, no patient has required
a surgical reintervention (Table 2).
Left ventricular ejection fraction increased significantly (P<.001) in both groups at 1 year compared with baseline preoperative
ventriculography (54%-61% in the OPCAB group and 53%-59% in the CABG with
cardiopulmonary bypass group).
Baseline preoperative SF-36 scores tended to be lower for OPCAB patients
and were significantly lower in the General Health Perception subscale and
the Mental Component scores. After adjusting for baseline, no significant
(P<.01) differences between groups were found
for any of the subscale or summary component scores of the SF-36 or the EuroQol
6 scores at any of the follow-up time points (Table 5). There was a trend toward a significant difference in the
Social Functioning subscale of the SF-36 at 1 year (P =
.049) favoring treatment with OPCAB.
There were 2 patients (both in the OPCAB group) with outlying hospital
costs due to severe noncardiac comorbidity (heparin-induced thrombocytopenia
and severe Clostridium difficile enterocolitis),
which were more than twice the next highest value for the group ($69 883
and $108 353 vs $33 348). Thus, for the primary analysis, hospital
costs for these patients were truncated back to the next highest observed
value ($33 348). Costs were truncated in the same manner for the patients
with the 2 highest observed hospital cost values in the CABG with cardiopulmonary
bypass group. Table 6 shows results
from the overall cost analysis based on truncated data. Mean index hospitalization
costs (hospital plus professional costs) were $2272 (95% CI, $755-$3732) lower
for the OPCAB group (P = .002). Without truncation
of outlying costs, average hospital costs for the index procedure were $18 236
for patients in the CABG with cardiopulmonary bypass group and $17 010
for patients in the OPCAB group (change in cost, $1226 [95% CI, −$1637
to $3575; P = .17). Costs associated with readmission
to hospital were similar between groups. The OPCAB treatment remained $1955
(95% CI, −$766 to $4727; P = .08) less expensive
on average at 1-year follow-up. Without truncating outlying costs, the difference
in 1-year costs was $1266 (95% CI, −$1613 to $3634; P = .16).
A preliminary report from the SMART trial19 documented
that the mean (SD) number of grafts performed per patient (OPCAB group: 3.39
[1.04]; CABG with cardiopulmonary bypass group: 3.40 [1.08]) and the number
of grafts performed per number of grafts intended (index of completeness of
revascularization) were similar between groups (OPCAB group: 1.00 [0.18];
CABG with cardiopulmonary bypass group: 1.01 [0.09]). This was the first time
that treatment with OPCAB achieved complete revascularization when applied
to unselected patients, avoiding the documented negative consequences of incomplete
Two previous randomized comparisons of OPCAB and CABG with cardiopulmonary
bypass have included postoperative angiography. Among low-risk patients, the
first study documented angiographic patency among a subset of approximately
25% of enrollees (158 grafts) and reported similar patency rates of 93% for
CABG with cardiopulmonary bypass and 91% for OPCAB at 1 year.18 The
second study9 reported 3-month angiographic
patency rates for 260 grafts among 82 of 103 randomized, low-risk patients.
The overall patency rate for grafts performed using conventional CABG was
significantly higher than for those performed using OPCAB (98% vs 88%; P = .002). The greatest difference between groups was for
patency of radial artery conduits (22/22 [100%] for CABG with cardiopulmonary
bypass compared with 26/34 [76%] for OPCAB; P = .01).
The 2 surgeons reporting these results had limited experience with OPCAB surgery,
having performed only 13% of their isolated coronary bypass procedures off-pump
during the preceding 2 years. However, each surgeon was obliged to perform
the procedure on each patient in the OPCAB group. Moreover, these surgeons
used a relatively low dose of heparin (150 units/kg) in their OPCAB patients
and did not supplement that dose intraoperatively.
All surgical procedures reported in the present trial were performed
by a single experienced OPCAB practitioner, who routinely performs more than
90% of his isolated coronary cases using OPCAB32 and
had personal experience in excess of 350 OPCAB cases prior to randomizing
patients for this trial. Heparin levels were supplemented at 30-minute intervals.
Angiographic results from the SMART trial, obtained in 184 of 197 patients
prior to hospital discharge (622 grafts) and 153 of 189 patients at 1 year
(511 grafts), demonstrate similar patency and completeness of revascularization
in both treatment groups. It is not possible to formally conclude that patency
was statistically equivalent between groups; more than 500 patients in each
treatment group would have been required to demonstrate such equivalence.
Nevertheless, these results suggest that application of OPCAB techniques to
unselected patients with multivessel disease need not lead to imprecise anastomoses,
nor to a decrement in graft patency.
Procoagulant activity may be increased after OPCAB.33,34 This
phenomenon has been invoked to explain the few isolated reports of diminished
vein graft patency after OPCAB.35,36 Importantly,
patency of saphenous vein grafts was similarly excellent between groups in
the present study. Patients in both groups of the SMART trial were treated
with perioperative and long-term aspirin; none were prescribed clopidogrel.
Nonetheless, awareness of a potential hypercoagulable state among OPCAB patients37 has prompted an institutional policy at Emory University
since January 2003 of treating OPCAB patients with postoperative clopidogrel
for 3 months.
Several previous studies have reported improved neuropsychological function
after treatment with OPCAB compared with CABG with cardiopulmonary bypass,38- 40 although this finding
has not been uniformly consistent.41 Results
from studies that compared quality-of-life measures after OPCAB and CABG with
cardiopulmonary bypass have also been inconsistent.41,42 While
the weight of peer-reviewed evidence may support the hypothesis that avoidance
of cardiopulmonary bypass improves neuropsychological and quality-of-life
outcomes after CABG surgery, this important area of investigation remains
limited by imprecise assessment tools43,44 and
logistical difficulties in performing perioperative testing.
Because economic considerations influence medical decision making, the
length of stay and cost associated with alternative methods of surgical coronary
revascularization are increasingly relevant. With all other acute and long-term
outcome measures either similar or superior for the OPCAB group, the decrease
in length of stay and cost may influence third-party payers and institutional
The present study has several important limitations. The performance
of all surgery by a single surgeon reduced surgical variability, thus making
the groups more comparable; but the generalizability of the findings to other
surgeons and health care systems has not been proven. No outpatient resource
use data was collected during the trial, and thus the evaluation of costs
in terms of hospitalization costs alone is limited. However, rehospitalization
data capture a large proportion of the costs associated with major adverse
events. If OPCAB is associated with lower morbidity compared with CABG with
cardiopulmonary bypass, this might be reflected in lower outpatient treatment
and medication use for patients receiving OPCAB. Under such circumstances,
results based on hospitalization costs alone would yield a conservative estimate
of the cost difference between groups.
Another important limitation is that 1-year follow-up angiography was
only obtained in 78% of the enrolled patients (81% of 1-year survivors). Although
there was no systematic difference between the 2 groups regarding lack of
1-year angiography, we cannot exclude the possibility of selection biases
affecting our findings. This type of limitation is inherent in trials that
require patients to undergo invasive follow-up procedures.
These results from the SMART trial demonstrate that OPCAB may provide
complete revascularization that is durable and cost-effective relative to
CABG with cardiopulmonary bypass when performed on unselected patients undergoing
elective, isolated CABG. A larger multicenter trial of OPCAB compared with
CABG with cardiopulmonary bypass is needed to evaluate the generalizability
of these results and to better clarify the role of OPCAB in the routine care
of patients with multivessel coronary artery disease.