JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO.

22, 2018

PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN

COLLEGE OF CARDIOLOGY FOUNDATION

Predictors and Outcomes of

Staged Versus One-Time
Multivessel Revascularization in
Multivessel Coronary Artery Disease
Insights From the VA CART Program

Peter T Hu, MD,a,b W. Schuyler Jones, MD,b,c Thomas J. Glorioso, MS,d,e Anna E. Barón, PHD,d,e
Gary K. Grunwald, PHD,d,e Stephen W. Waldo, MD,d,f Thomas M. Maddox, MD, MSC,g Mladen Vidovich, MD,h,i
Subhash Banerjee, MD,j,k Sunil V. Rao, MDb,c,l

ABSTRACT

OBJECTIVES The aim of this study was to determine predictors and outcomes associated with staged percutaneous
coronary intervention (PCI) versus one-time multivessel revascularization (OTMVR) in patients with multivessel coronary
artery disease.

BACKGROUND Prior observational studies have not evaluated predictors and outcomes of staged PCI versus OTMVR in
a heterogenous population of patients with multivessel coronary artery disease who undergo multivessel
revascularization.

METHODS Data from the Veterans Affairs (VA) CART (Clinical Assessment, Reporting, and Tracking) Program were used
to evaluate patients who underwent PCI of >2 vessels between October 1, 2007, and September 3, 2014. Associations
between individual factors and the decision to perform staged PCI were assessed. Additionally, the impact of measured
patient and procedural factors, site factors, and unmeasured site factors on the decision to perform staged PCI was
compared. Cox proportional hazards models were used to determine the association between staged PCI and mortality.

RESULTS A total of 7,599 patients at 61 sites were included. The decision to perform staged PCI was driven by pro-
cedural characteristics and unmeasured site factors. Staged PCI was associated with lower risk-adjusted mortality
compared with OTMVR (adjusted hazard ratio [HR]: 0.78; 95% confidence interval [CI]: 0.72 to 0.84; p < 0.01). This
mortality benefit was observed among the ST-segment elevation myocardial infarction (HR: 0.31; 95% CI: 0.21 to 0.47;
p < 0.01), non–ST-segment elevation myocardial infarction (HR: 0.74; 95% CI: 0.64 to 0.87; p < 0.01), unstable angina
(HR: 0.75; 95% CI: 0.64 to 0.89; p < 0.01) and stable angina (HR: 0.88; 95% CI: 0.77 to 1.00; p ¼ 0.05) groups.

CONCLUSIONS The decision to pursue staged PCI was driven by procedural characteristics and unmeasured site
variation and was associated with lower mortality compared with OTMVR. After adjustment, there was an association
between staged PCI and reduced mortality. Given the observational nature of these findings, a randomized trial
comparing the 2 is needed to guide practice. (J Am Coll Cardiol Intv 2018;11:2265–73) Published by Elsevier on behalf of
the American College of Cardiology Foundation.

From the aDepartment of Cardiology, Cleveland Clinic, Cleveland, Ohio; bDepartment of Medicine, Duke University Medical
Center, Durham, North Carolina; cDuke Clinical Research Institute, Duke University Medical Center, Durham, North Carolina;
d
Denver Veterans Affairs Medical Center, Denver, Colorado; eColorado School of Public Health, University of Colorado Anschutz
Medical Campus, Aurora, Colorado; fSection of Cardiology, University of Colorado School of Medicine, Aurora, Colorado; gDivision
of Cardiology, Washington University, St. Louis, Missouri; hUniversity of Illinois College of Medicine, Chicago, Illinois; iJesse
Brown Veterans Affairs Medical Center, Chicago, Illinois; jUniversity of Texas Southwestern Medical Center, Dallas, Texas;
k
Veterans Affairs North Texas Health Care System, Dallas, Texas; and the lDurham Veterans Affairs Medical Center, Durham,
North Carolina. Dr. Jones has received research funding from the Agency for Healthcare Research and Quality, the American Heart
Association, AstraZeneca, Bristol-Myers Squibb, Daiichi-Sankyo, and the Patient-Centered Outcomes Research Institute. Dr. Waldo

ISSN 1936-8798/$36.00 https://doi.org/10.1016/j.jcin.2018.07.055

2266 Hu et al. JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018

Staged Versus One-Time Multivessel Revascularization NOVEMBER 26, 2018:2265–73

M
ABBREVIATIONS ultivessel coronary artery disease associated outcomes of MVPCI across the spectrum of
AND ACRONYMS (MVCAD) is present in approxi- clinical presentation remain unclear, and second,
mately two-thirds of patients observational studies are subject to multiple biases,
ACS = acute coronary
syndrome(s)
who require revascularization (1). Two including unmeasured confounding. To date, there
options exist when performing multi- are insufficient observational data and limited ran-
CI = confidence interval
vessel percutaneous coronary intervention domized controlled trials to provide recommenda-
HR = hazard ratio
(MVPCI): staged percutaneous coronary tions on the optimal timing of MVPCI (8).
MVCAD = multivessel coronary
artery disease
intervention (PCI) or one-time multivessel Using standardized electronic health system data
revascularization (OTMVR). In staged PCI, from the national Veterans Affairs (VA) CART-CL
MVPCI = multivessel
percutaneous coronary patients often receive revascularization of a (Clinical Assessment, Reporting, and Tracking Sys-
intervention culprit (thrombotic) artery or the most severe tem for Catheterization Laboratories) Program, we
NSTEMI = non–ST-segment stenosis, followed by a planned revasculari- sought to determine: 1) factors associated with the
elevation myocardial infarction
zation of a nonculprit artery or arteries or decision to perform staged PCI or OTMVR among
OR = odds ratio less severe or complex stenosis at a later patients who underwent MVPCI, including unmea-
OTMVR = one-time multivessel time. This is in contrast to OTMVR, in which sured site variation; and 2) the association between
revascularization
more than 1 diseased lesion is revascularized staged PCI or OTMVR and clinical outcomes (death,
PCI = percutaneous coronary
at the time of the index procedure. Patients rehospitalization, transfusion) (9).
intervention
with MVCAD who undergo MVPCI constitute
REM = reference effect
a heterogenous patient population. The deci-
measure METHODS
sion to perform staged PCI or OTMVR may be
STEMI = ST-segment elevation
myocardial infarction related to many factors, including patient-
STUDY SAMPLE. The VA CART Program is a national
VA = Veterans Affairs specific factors (e.g., chronic kidney disease,
VA clinical quality initiative that was launched in
indication for PCI such as ST-segment eleva-
2005 and has been described previously (10). A key
tion myocardial infarction [STEMI], non–ST-segment
feature of the CART Program is standardized data
elevation myocardial infarction [NSTEMI], unstable
capture and reporting across all VA catheterization
angina, or stable angina), physician- or hospital-
laboratories through a clinical software application.
specific factors, and procedural characteristics (e.g.,
The software is embedded in the VA electronic health
anatomic complexity, disease burden, presence of
record and allows providers to enter patient and
chronic total occlusion, contrast load, radiation expo-
procedural information (pre-procedural assessment,
sure). Each of these factors plays an important role in
cardiac catheterization, and PCI) as part of routine
the decision-making process in MVPCI. For example,
clinical work flow. The application is integrated with
patients with stable angina and chronic total occlu-
the VA’s patient electronic health record, and data
sions who undergo MVPCI are more likely to undergo
elements automatically populate a clinical note.
staged PCI (2). Furthermore, among interventional
These data are combined with longitudinal VA
cardiologists, renal function is the strongest factor
administrative data in a clinical data repository to
in the decision to pursue staged PCI (3).
support the quality assessment, quality improve-
SEE PAGE 2274 ment, and clinical research missions of the CART
Program. Data elements conform to the definitions
In patients with STEMI, several observational and standards of the American College of Cardiol-
studies have suggested that multivessel staged PCI ogy’s National Cardiovascular Data Registry (9,11).
may be associated with lower mortality compared Cardiac catheterization reports generated by the VA
with OTMVR (4–7). Observational data comparing CART Program have been shown to demonstrate
staged versus OTMVR in those with NSTEMI are excellent data validity and completeness (10). This
lacking. Two issues hinder the use of these data to study was approved by the Institutional Review
guide practice. First, the optimal timing and Board at the Denver VA Medical Center and the

receives unrelated research funding to the Denver Research Institute from Abiomed, Cardiovascular Systems Inc., and Merck
Pharmaceuticals. Dr. Maddox was the former director of the VA CART Program. Dr. Vidovich has received research funding from
Boston Scientific, Biosensors International, and Sanofi; and is a consultant for Boston Scientific and Abbott and serves on the
advisory board for Merit. Dr. Banerjee has received honoraria from Medtronic, Gore, and Cardiovascular Systems Inc., and
research funding from Boston Scientific and Merck. All other authors have reported that they have no relationships relevant to the
contents of this paper to disclose.

Manuscript received June 5, 2018; revised manuscript received July 24, 2018, accepted July 31, 2018.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018
NOVEMBER 26, 2018:2265–73
Durham VA Medical Center, with a waiver of the
requirement to obtain informed consent.
STUDY POPULATION. Patients included in the anal-
ysis were those who had MVCAD and underwent PCI to
2 or more coronary artery distributions at a VA cathe-
terization laboratory between October 1, 2007, and
September 30, 2014, and had their records collected in
CART. Patients were excluded from analyses if they
had left main coronary artery treatment, presented
with cardiogenic shock, underwent PCI in the previous
60 days either inside or outside of the VA, underwent
more than 1 staged procedure, or were at a VA cathe-
terization laboratory that performed no staged PCI or
had performed 10 or fewer PCIs overall. If a patient
underwent more than 1 MVPCI over this time frame,
only the first PCI was used. Patients with missing data
were not included in the analyses (Figure 1).
DEFINITIONS AND OUTCOMES. O T M V R . OTMVR
was defined as PCI of >2 diseased coronary artery
distributions at the time of an index procedure. A
diseased left anterior descending, left circumflex, or
right coronary artery was defined by a segment with
>70% stenosis in a main branch or side branch of an
epicardial artery.
S t a g e d P C I . To be included in the staged PCI group,
a subsequent PCI of a diseased coronary artery had to
be performed within 60 days of an index procedure.
The second procedure could not have treated a
segment that had been treated during the index pro-
cedure. The subsequent procedure could not have a
PCI status of emergent, urgent, or salvage. It also
could not have an indication of STEMI or NSTEMI. If
comments in CART-CL indicated that a patient should
be scheduled for a staged procedure, the patient was
included as part of the staged PCI group using an
intention-to-treat approach.
O u t c o m e s . The primary efficacy outcome was all-
cause death through latest follow-up. Secondary
outcomes included all-cause rehospitalization and
rehospitalization for myocardial infarction through
latest follow-up. These were time-to-event outcomes
using the latest discharge date from the index or
staged procedure as a baseline date. Follow-up data
were available through September 30, 2015, and thus
patients with no events were censored as of this date.
For outcomes other than mortality, patients were
censored at the date of death or the last date of follow-
up. The primary safety outcome was the rate of blood
transfusions within 72 h after both the index PCI and
subsequent PCI if a patient had a staged procedure.
PATIENT COMORBIDITIES. With the exception of
body mass index and baseline renal function, Inter-
national Classification of Diseases-9th Revision codes
Hu et al. 2267
Staged Versus One-Time Multivessel Revascularization
in prior inpatient, outpatient, and CART records were
used to determine medical comorbidities. Two
outpatient records or 1 inpatient record with the In-
ternational Classification of Diseases-9th Revision
code of interest were needed to confirm a comorbid
condition.
PROCEDURAL CHARACTERISTICS. Information from
either the index procedure or a combination of the
index and staged procedure was used for the proce-
dural characteristic variables. Number of stents used
and number of diseased vessels treated were summed
over both procedures if a patient underwent multiple
PCIs. PCI status, indication, and access site (radial vs.
femoral) were used from the index procedure. If a
patient had multiple access sites listed for the index
procedure, the second access site was used.
SITE CHARACTERISTICS. Site regional status was
based on U.S. census regions (West, Midwest, South,
and Northeast). Certain patient-level characteristics
that may better describe the catheterization labora-
tory site, such as racial breakdown and urban or rural
status, were summarized at the site level by taking
the overall proportion of patients in the cohort with
the characteristic of interest at the site.
STATISTICAL ANALYSIS. All data were reviewed for
missing values and quality. Some patients were
missing values for body mass index (1.2%) and urban
or rural status (1.1%), and a single imputation was
performed simulating values on the basis of associa-
tions with observed data; 449 patients with other
missing values were excluded because of the diffi-
culty in correctly imputing values.
A mixed logistic regression model with a random
intercept for site was estimated using the glmmML
package in R version 3.2.5 (R Foundation for Statistical
Computing, Vienna, Austria). Associations of individ-
ual measured patient and site factors with staged PCI
versus OTMVR, adjusted for other patient, procedure,
and site factors, were examined using this model. We
used reference effect measure (REM) methods to
explore the extent to which unmeasured site variation
contributed to the decision to use staged PCI relative
to sets of measured factors (12). A REM value is an odds
ratio (OR) comparing a patient at a given risk percen-
tile with a median-risk patient. For variation between
sites not accounted for by measured patient or site
characteristics (unmeasured site variation), this is
based on the estimated normal distribution of the site
random effects. For variation due to measured de-
mographics, comorbidities, procedure, or site charac-
teristics, REM is based on the corresponding empirical
distributions of patient risk scores. For each
2268 Hu et al.
Staged Versus One-Time Multivessel Revascularization
F I G U R E 1 Consolidated Standards of Reporting Trials Diagram
ACS ¼ acute coronary syndrome(s); NSTEMI ¼ non–ST-segment elevation myocardial
infarction; OTMVR ¼ one-time multivessel revascularization; PCI ¼ percutaneous coro-
nary intervention; STEMI ¼ ST-segment elevation myocardial infarction.
distribution, the OR for staged PCI for a low-risk (2.5th
percentile) and a high-risk (97.5th percentile) patient
relative to a median-risk patient formed 95% REM
ranges. Wider ranges indicated greater differences in
“risk” or probability of having a staged procedure on
the basis of the factors of interest and thus suggested a
larger contribution to the procedural decision.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018
NOVEMBER 26, 2018:2265–73
A propensity score approach using inverse proba-
bility of treatment weighting assessed the association
between staged PCI and efficacy outcomes balancing
for differences in patient risk. A hybrid method using
both parametric and nonparametric models was used
to produce propensity scores and weights (13). Vari-
ables in the propensity score models included patient
demographics (age, sex, race, urban vs. rural),
comorbidities (body mass index, congestive heart
failure, chronic kidney disease, chronic obstructive
pulmonary disease, diabetes, hyperlipidemia, hyper-
tension, peripheral artery disease, prior myocardial
infarction, prior coronary artery bypass graft surgery,
prior PCI, prior stroke or transient ischemic attack,
tobacco use), procedural characteristics (use of at
least 1 bare-metal stent, number of diseased vessels,
number of stents, indication, status), site character-
istics (proportion of urban patients, proportion of
white patients, coronary artery bypass graft surgery
on-site), and time trend. A Cox proportional hazards
model using these weights and a frailty term for site
was then estimated using the coxme package, and we
report the hazard ratio (HR) for staged PCI versus
OTMVR. Diagnostics from the propensity-weighted
cohort are provided in Online Table 1 and Online
Figure 1. Sensitivity analyses incorporating unmea-
sured site variation into the propensity score also
produced similar results (14). Additionally, sensitivity
analyses for unmeasured confounding examined
changes in the association between OTMVR versus
staged PCI and mortality with the inclusion of a
hypothetical unmeasured confounder with assumed
prevalence in each group and strength of association
with the outcome (15) (Online Table 2).
Because of the small number of events in the safety
outcome, an unadjusted chi-square test compared the
proportion of patients with blood transfusions by
staged PCI versus OTMVR. Results were restricted to
the 7,578 of 7,599 patients who survived to the 72-h
cutoff after their last procedure. Finally, subgroup
analyses were performed using the previously
described propensity score weighting approach
among patients presenting with acute coronary syn-
drome (ACS), non-ACS, STEMI, NSTEMI, unstable
angina, and stable angina. All analyses were per-
formed by the VA CART analytic center at the Denver
VA Medical Center using R version 3.2.5.
RESULTS
STUDY POPULATION. A total of 10,498 patients un-
derwent MVPCI at 67 sites between October 1, 2007,
and September 30, 2014. After exclusion criteria were
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018 Hu et al. 2269
NOVEMBER 26, 2018:2265–73 Staged Versus One-Time Multivessel Revascularization

applied, the final study sample included 7,599 pa-

T A B L E 1 Baseline Patient Characteristics of Multivessel Percutaneous Coronary
tients at 61 sites (Figure 1). Of the patients in the Intervention Patients (67 Percutaneous Coronary Intervention Sites Between
analysis cohort, 2,743 (36.1%) underwent staged PCI, October 1, 2007, and September 30, 2014): Staged Percutaneous Coronary Intervention
while 4,856 (63.9%) underwent OTMVR. Versus One-Time Multivessel Revascularization

A summary of baseline patient and procedure All Staged PCI OTMVR

characteristics grouped by staged PCI versus OTMVR (N ¼ 7,599) (n ¼ 2,743) (n ¼ 4,856) p Value

is shown in Table 1. Including both the index and Demographics

subsequent staged procedures, patients with MVCAD Age (yrs) 65.3  9.0 65.2  8.9 65.3  9.1 0.49

who underwent staged PCI had higher rates of Sex (male) 0.987 (7,502) 0.989 (2,713) 0.986 (4,789) 0.34
Race (white) 0.848 (6,441) 0.859 (2,356) 0.841 (4,085) 0.04
treated 3-vessel disease (12.3% vs. 4.4%; p < 0.01),
Hispanic 0.044 (338) 0.044 (120) 0.045 (218) 0.86
STEMI presentation (8.1% vs. 3.0%; p < 0.01), and a
Urban 0.581 (4,415) 0.575 (1,578) 0.584 (2,837) 0.46
PCI status of emergent or salvage (8.0% vs. 3.4%;
Comorbidities
p < 0.01) compared with patients with OTMVR. The
BMI (kg/m2) 30.3  5.6 30.2  5.5 30.3  5.7 0.76
mean total number of stents placed during index PCI
CHF 0.231 (1,752) 0.212 (582) 0.241 (1,170) <0.01
and subsequent PCI for those who underwent staged
CKD 0.202 (1,538) 0.209 (572) 0.199 (966) 0.33
procedures (3.2) was greater than that for patients COPD 0.208 (1,583) 0.199 (546) 0.214 (1,037) 0.14
who underwent OTMVR (2.6) (p < 0.01). Baseline Diabetes 0.501 (3,809) 0.486 (1,334) 0.51 (2,475) 0.05
mean glomerular filtration rate did not differ GFR (ml/min/1.73 m2) 75.0  25.4 74.7  25.1 75.2  25.6 0.38
between those who underwent staged PCI versus Hyperlipidemia 0.898 (6,822) 0.883 (2,422) 0.906 (4,400) <0.01
OTMVR (mean glomerular filtration rate 74.7 vs. Hypertension 0.902 (6,857) 0.886 (2,431) 0.911 (4,426) <0.01
75.2 ml/min/1.73 m 2; p ¼ 0.38). Compared with PAD 0.194 (1,473) 0.184 (504) 0.2 (969) 0.10
OTMVR, patients undergoing staged PCI had higher Prior CABG 0.134 (1,020) 0.131 (360) 0.136 (660) 0.59
rates of atherectomy use (14.4% vs. 8.3%; p < 0.01), Prior MI 0.343 (2,609) 0.309 (847) 0.363 (1,762) <0.01
treated chronic total occlusions (16.4% vs. 8.8%; Prior PCI 0.357 (2,715) 0.331 (908) 0.372 (1,807) <0.01

p < 0.01) and treated calcified lesions (39.4% vs. Prior stroke/TIA 0.085 (649) 0.076 (209) 0.091 (440) 0.03

31.3%; p < 0.01) during either the index or staged Tobacco 0.635 (4,823) 0.601 (1,649) 0.654 (3,174) <0.01
Procedural characteristics
procedure.
Access (radial) 0.141 (1,072) 0.126 (345) 0.15 (727) <0.01
FACTORS ASSOCIATED WITH STAGED PCI. Stent type (BMS) 0.151 (1,144) 0.15 (412) 0.151 (732) 0.98
Table 2 shows adjusted associations between indi- Number of stents 2.8  1.3 3.2  1.5 2.6  1.1 <0.01
vidual factors and staged PCI versus OTMVR. With Treated vessels <0.01
respect to baseline patient characteristics, the only 2 0.927 (7,047) 0.877 (2,405) 0.956 (4,642)
comorbidity associated with staged PCI was chronic 3 0.073 (552) 0.123 (338) 0.044 (214)
kidney disease (adjusted OR: 1.31; 95% confidence Indication <0.01
interval [CI]: 1.14 to 1.52; p < 0.01). Procedural char- STEMI 0.048 (367) 0.081 (222) 0.03 (145)

acteristics associated with staged PCI included NSTEMI 0.213 (1,618) 0.213 (585) 0.213 (1,033)

3-vessel disease treated versus 2-vessel disease Stable angina/chest 0.418 (3,180) 0.409 (1,123) 0.424 (2,057)
pain/asymptomatic
(adjusted OR: 2.17; 95% CI: 1.77 to 2.68; p < 0.01) and Unstable angina/ACS 0.262 (1,991) 0.241 (661) 0.274 (1,330)
STEMI presentation (adjusted OR: 2.48; 95% CI: 1.74 Other/valve disease 0.058 (443) 0.055 (152) 0.06 (291)
to 3.52; p < 0.01). CTO treated 0.116 (878) 0.164 (450) 0.088 (428) <0.01
In Figure 2, we observe a wide 95% REM range for Calcified treated 0.342 (2,600) 0.394 (1,080) 0.313 (1,520) <0.01
unmeasured site variation (OR: 0.20 to 5.08), which Atherectomy 0.105 (797) 0.144 (395) 0.083 (402) <0.01
was comparable in width with the 95% REM range for Distal protection <0.001 (6) 0.001 (3) 0.001 (3) 0.78

procedural characteristics (OR: 0.44 to 6.17) and Embolic protection 0.001 (8) 0.001 (2) 0.001 (6) 0.78

exceeded demographics (OR: 0.79 to 1.11), comorbid- Status <0.01

ities (OR: 0.69 to 1.39), and measured site character- Elective 0.642 (4,878) 0.632 (1,733) 0.648 (3,145)
Urgent 0.307 (2,336) 0.288 (791) 0.318 (1,545)
istics (OR: 0.44 to 2.14). On the basis of unmeasured
Emergent/salvage 0.051 (385) 0.08 (219) 0.034 (166)
site variation, a patient at a low-“risk” site (2.5th
percentile) has odds of a staged procedure 0.20 times Values are mean  SD or proportion (n).
lower compared with a patient a median-“risk” site, ACS ¼ acute coronary syndrome(s); BMI ¼ body mass index; BMS ¼ bare-metal stent(s); CABG ¼ coronary
artery bypass graft; CHF ¼ congestive heart failure; CKD ¼ chronic kidney disease; COPD ¼ chronic obstructive
while a patient at a high-“risk” site (97.5th percentile) pulmonary disease; CTO ¼ chronic total occlusion; GFR ¼ glomerular filtration rate; MI ¼ myocardial infarction;
has odds 5.08 times greater. The wide ranges show NSTEMI ¼ non–ST-segment elevation myocardial infarction; OTMVR ¼ one-time multivessel revascularization;
PAD ¼ peripheral artery disease; PCI ¼ percutaneous coronary intervention; STEMI ¼ ST-segment elevation
large differences in the likelihood of receiving staged myocardial infarction; TIA ¼ transient ischemic attack.
PCI due to unmeasured site variation and procedural
2270 Hu et al. JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018

Staged Versus One-Time Multivessel Revascularization NOVEMBER 26, 2018:2265–73

with a lower risk for death (HR: 0.78; 95% CI: 0.72 to
T A B L E 2 Predictors of Staged Percutaneous Coronary
Intervention Versus One-Time Multivessel Revascularization
0.84; p < 0.01). For the secondary outcomes of all-
cause rehospitalization and rehospitalization for
OR (95% CI) p Value
myocardial infarction after propensity adjustment,
Demographics
there were no statistically significant differences
Urban 0.91 (0.81–1.03) 0.13
(Table 3). Sensitivity analyses (Online Table 2) show
Hispanic 0.91 (0.69–1.21) 0.52
the effect of an unmeasured confounder with varying
Age (10 yrs) 0.98 (0.92–1.05) 0.63
prevalence and its association with death.
Race (white) 1.13 (0.96–1.33) 0.14
Sex (male) 1.25 (0.76–2.06) 0.38 ASSOCIATION BETWEEN STAGED PCI AND TRANS-
Comorbidities FUSION. The proportion of patients receiving blood
Prior stroke/TIA 0.87 (0.71–1.06) 0.15 transfusion within 72 h of both index PCI and subse-
CHF 0.87 (0.76–1.01) 0.06
quent PCI if a patient underwent a staged procedure
Prior MI 0.88 (0.77–1.00) 0.04
was not significantly different between the staged PCI
Hypertension 0.88 (0.73–1.06) 0.19
(1.4%) and OTMVR (1.3%) groups (p ¼ 0.96).
Diabetes 0.90 (0.80–1.01) 0.07
Tobacco use 0.91 (0.81–1.02) 0.10
SUBGROUP ANALYSIS. Similar HRs for mortality

COPD 0.91 (0.80–1.05) 0.20

were observed when comparing staged PCI with
Prior PCI 0.96 (0.85–1.09) 0.57 OTMVR among patients who presented with ACS
PAD 0.97 (0.84–1.12) 0.68 (adjusted HR: 0.73; 95% CI: 0.66 to 0.81; p < 0.01).
Hyperlipidemia 1.00 (0.82–1.20) 0.99 This significant relationship was consistent in those
BMI 1.01 (1.00–1.02) 0.24 who presented with STEMI (HR: 0.31; 95% CI: 0.21 to
Prior CABG 1.17 (0.99–1.37) 0.07 0.47; p < 0.01), NSTEMI (HR: 0.74; 95% CI: 0.64 to
CKD 1.31 (1.14–1.52) <0.01 0.87; p < 0.01), unstable angina (HR: 0.75; 95% CI:
Procedural characteristics 0.64 to 0.89; p < 0.01), or stable angina (HR: 0.88;
BMS (at least 1) 0.97 (0.83–1.14) 0.72 95% CI: 0.77 to 1.00; p ¼ 0.048) (Table 4).
Urgent vs. elective 1.00 (0.86–1.15) 0.96
Unstable angina vs. stable angina 1.02 (0.88–1.18) 0.82
DISCUSSION
Other vs. stable angina 1.15 (0.90–1.46) 0.26
NSTEMI vs. stable angina 1.16 (0.98–1.38) 0.09
Radial access 1.34 (1.10–1.62) <0.01
In this contemporary study of patients with MVCAD
Emergent vs. elective 1.45 (1.03–2.04) 0.03 undergoing staged PCI or OTMVR, we report several
Number of stents 1.52 (1.45–1.59) <0.01 key findings: 1) although some patient and procedure
3 vs. 2 vessels treated 2.17 (1.77–2.68) <0.01 factors are associated with staged PCI, site variation
STEMI vs. stable angina 2.48 (1.74–3.52) <0.01 due to unmeasured factors had a substantial impact
Site characteristics on the decision to perform staged PCI; 2) after
West vs. South 0.37 (0.18–0.77) 0.01 adjustment for differences in patient and procedure
Northeast vs. South 0.44 (0.20–0.98) 0.04 characteristics, staged PCI was associated with lower
CABG surgical on-site 0.95 (0.56–1.63) 0.86 risk for death compared with OTMVR overall and
Site (n ¼ 100) 0.99 (0.81–1.19) 0.88 across all subgroups; and 3) overall the rates of
Proportion urban (10%) 1.12 (0.99–1.28) 0.08 bleeding with staged PCI and OTMVR were low and
Midwest vs. South 1.14 (0.64–2.05) 0.65 not significantly different between groups. However,
Proportion white (10%) 1.28 (1.02–1.61) 0.03
given the influence of unmeasured site variation and
CI ¼ confidence interval; OR ¼ odds ratio; other abbreviations as in Table 1.
potential unmeasured confounding on these re-
lationships, further exploration of which factors (e.g.,
patient preference, physician preference) is war-
characteristics, indicating they are driving the pro- ranted, and a prospective randomized trial is needed
cedural decision. to guide clinical practice in this population.
ASSOCIATION BETWEEN STAGED PCI AND To date, there are no completed randomized
EFFICACY OUTCOMES. The median time of follow- trials in patients with MVCAD comparing staged
up was 43.7 months for the primary outcome of versus complete revascularization. Prior studies
all-cause death. In unadjusted analysis, staged PCI comparing single-vessel PCI versus MVPCI have
was associated with a lower risk for death compared been conducted largely in patients presenting with
with OTMVR (HR: 0.76; 95% CI: 0.68 to 0.85; p < STEMI who have MVCAD. The results of the PRAMI,
0.01). After adjustment including applying the pro- CvLPRIT, and DANAMI-3-PRIMULTI randomized
pensity scores weights, staged PCI was still associated controlled trials found significant benefit with
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018 Hu et al. 2271
NOVEMBER 26, 2018:2265–73 Staged Versus One-Time Multivessel Revascularization

complete revascularization compared with culprit

F I G U R E 2 Reference Effect Measure Plot of Risk Distributions for
artery PCI alone (16–18). The results of these trials Combined Demographic, Comorbidity, Site, and Procedural Characteristics and
changed treatment recommendations for complete Unmeasured Variation Across Sites, Staged Percutaneous Coronary Intervention
revascularization at the time of primary PCI in Versus One-Time Multivessel Revascularization

patients with STEMI without hemodynamic compro-

mise from a Class IIIC to a Class IIB recommendation
(8). The subsequent COMPARE ACUTE trial again
showed improved overall outcomes in patients with
STEMI who underwent complete revascularization
versus infarct artery–only revascularization in the
acute setting. This was driven by a decreased rate of
future revascularization (19). The question of staged
PCI after culprit artery revascularization in patients
with STEMI is being evaluated in the ongoing COM-
PLETE trial. Notably PRAMI, CvLPRIT, DANAMI-3-
PRIMULTI, and COMPARE ACUTE excluded patients
with STEMI with cardiogenic shock. The recently
published CULPRIT-SHOCK trial showed worse out-
comes in patients with STEMI with cardiogenic shock
who underwent culprit artery–only and non–culprit
artery–only revascularization, suggesting that there
is still a role for culprit artery–only revascularization
in a select group of patients (20). In patients with
STEMI, the COMPLETE trial will likely confirm a role
for staged multivessel revascularization following
culprit artery revascularization. However, the optimal
timing of non–culprit artery revascularization in pa-
tients with STEMI remains unknown without a direct
comparison between staged PCI and OTMVR.
An aspect of our analysis that is interesting is the
finding that staged PCI is associated with lower long-
term mortality versus OTMVR. In patient with STEMIs
in particular, staged PCI had a strong association with
mortality (HR: 0.31; 95% CI: 0.21 to 0.47; p < 0.01).
The mortality benefit we observed is consistent with
prior meta-analyses (21,22). The increased mortality
in patients with STEMI who undergo culprit and non– Percentiles from distributions assessing the variation in the “risk” of staged percutaneous
culprit artery revascularization in the acute period coronary intervention (PCI) relative to a median-risk patient for groups of factors and
may be attributed to a proinflammatory state leading unmeasured site variation. Wider ranges indicate larger differences in the likelihood to stage

to an increased risk for in-stent thrombosis with PCI
of nonculprit vessels, higher contrast load leading to
contrast-induced nephropathy, which has
associated with increased mortality, or greater risk for
procedural complications with increased procedure
times and/or intervening on nonculprit vessels
without optimal evaluation (e.g., percentage diam-
eter stenosis, fractional flow reverse, or intravascular
ultrasound) (23). In those without STEMI,
observed a positive yet weaker association with lower
mortality in the NSTEMI (HR: 0.74; 95% CI: 0.64 to
0.87; p < 0.01), unstable angina (HR: 0.75; 95% CI:
0.64 to 0.89; p < 0.01), and stable angina (HR: 0.88;
95% CI: 0.77 to 1.00; p ¼ 0.05) groups. This trend
PCI on the basis of the group of factors, suggesting that they contribute more to the
procedural decision. The width of the 95% reference effect measure (REM) range for
unmeasured site variation (odds ratio [OR]: 0.20 to 5.08) exceeded that of demographics
been (OR: 0.79 to 1.11), comorbidities (OR: 0.69 to 1.39), and site characteristics (OR: 0.44 to
2.14) and was comparable in width to procedural characteristics (OR: 0.44 to 6.17).
toward insignificance in the stable angina group from
the STEMI, NSTEMI, and unstable angina groups is
we likely attributed to a lesser degree of acuity and less
of an inflammatory state as well as procedural related
complications. This is the first analysis we are aware
of to find this association across subgroups.
To assess whether a second procedure in selected
patients would be associated with an increase in
2272 Hu et al. JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018

Staged Versus One-Time Multivessel Revascularization NOVEMBER 26, 2018:2265–73

In addition to the significant impact of unmeasured

T A B L E 3 Adjusted Hazard Ratios for the Primary and
Secondary Efficacy Endpoints for Staged Percutaneous Coronary
site variation, we found that a composite of proce-
Intervention Versus One-Time Multivessel Revascularization dural factors affected the decision to perform staged
PCI. Furthermore, individual patient factors, such as
Hazard Ratio 95% CI p Value
chronic kidney disease, and procedural factors, such
Death 0.78 (0.72–0.84) <0.01
as presentation with STEMI and 3-vessel coronary
Rehospitalization 0.98 (0.94–1.03) 0.46
disease, were associated with the use of staged PCI. In
Rehospitalization for MI 0.90 (0.79–1.02) 0.09
the absence of clinical guidelines on the optimal
Abbreviations as in Tables 1 and 2. timing of MVPCI, the decision to pursue staged PCI
varies widely among cardiologists. This was observed
in a survey performed by the American College of
adverse procedural complications such as bleeding, Cardiology (3). There is general agreement among
we examined the association between staged PCI and cardiologists that renal function, lesion complexity,
blood transfusions. We found that there was no sig- and presence of ACS are important factors in the
nificant difference in post-PCI blood transfusion be- decision to pursue staging (3).
tween the staged PCI and OTMVR groups. This is STUDY LIMITATIONS. First, although we controlled
consistent with prior results in patients with MVCAD for known confounders and used propensity matching
who underwent revascularization, suggesting that to ensure that the staged PCI and OTMVR groups were
the decision to stage PCI is not associated with an similar, the analysis is observational and subject to
adverse safety profile (4,17,24). unmeasured confounding. Second, it is possible that
Despite the robustness of our results, they are patients who underwent index PCI with plans for
observational in nature. Unlike prior studies, we staged PCI may not have returned for the subsequent
documented the degree to which unmeasured site procedure. We accounted for this by using an
variation influenced the decision to pursue staged intention-to-treat analysis; however, if a planned
PCI using the REM approach. Compared with de- approach was not mentioned in CART-CL, some pa-
mographics and comorbidities, procedural charac- tients may have been misclassified as OTMVR or
teristics and unmeasured site variation contributed excluded entirely if only a single vessel was treated,
more to the decision for staged PCI versus OTMVR as leading to survivor bias. Similarly, it is possible that
shown by the wide 95% REM ranges for these fac- some patients who were to undergo planned OTMVR
tors. These data indicate that local practice patterns subsequently underwent staged PCI because of pro-
may play a large role in this process. Factors that cedural difficulties during the index case and were not
may also contribute to unmeasured site variation included as part of the OTMVR group. Third, our study
include physician preference, technical skill, differ- sample was largely male, given the demographic
ences in degree of illness at patient presentation makeup of the veteran patient population. Similarly,
across sites, and differences in systems-based prac- the results of our study may not be generalizable to a
tices. This substantial variation in clinical practice nonveteran population. Finally, we were unable to
highlights the need for implementation of guideline assess the degree of angiographic complexity in our
directed care. study sample, and those details may account for some
of the unmeasured confounding in the propensity to
undergo staged PCI versus OTMVR.
T A B L E 4 Mortality Rates (Hazard Ratios) After Propensity CONCLUSIONS
Matching for Staged Percutaneous Coronary Intervention
Compared With One-Time Multivessel Revascularization
by Subgroup In this analysis of patients with MVCAD undergoing
PCI, the decision to perform staged PCI was driven by
Total n Hazard Ratio 95% CI p Value
specific patient characteristics but more by procedural
ACS 3,976 0.73 (0.66–0.81) <0.01
characteristics and unmeasured site variation, and
Non-ACS 3,623 0.87 (0.77–0.97) 0.02
staged PCI was associated with a significant mortality
STEMI 367 0.31 (0.21–0.47) <0.01
benefit compared with OTMVR.
NSTEMI 1,618 0.74 (0.64–0.87) <0.01
Unstable angina 1,991 0.75 (0.64–0.89) <0.01
ADDRESS FOR CORRESPONDENCE: Dr. Sunil V. Rao,
Stable angina 3,180 0.88 (0.77–1.00) 0.048
Durham VA Medical Center, 508 Fulton Street (111A),
Abbreviations as in Tables 1 and 2.
Durham, North Carolina 27705. E-mail: sunil.rao@
duke.edu.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 11, NO. 22, 2018
NOVEMBER 26, 2018:2265–73
PERSPECTIVES
WHAT IS KNOWN? In patients with STEMI, several
observational studies have suggested that multivessel
staged PCI may be associated with lower mortality
compared with OTMVR.
WHAT IS NEW? This study incorporates a heterogenous
patient population with STEMI, NSTEMI, unstable angina,
and stable angina who underwent either staged PCI or
REFERENCES
1. Pocock SJ, Henderson RA, Rickards AF, et al.
Meta-analysis of randomised trials comparing
coronary angioplasty with bypass surgery. Lancet
1995;346:1184–9.
2. Brilakis ES, Banerjee S, Karmpaliotis D, et al.
Procedural outcomes of chronic total occlusion
percutaneous coronary intervention: a report from
the NCDR (National Cardiovascular Data Registry).
J Am Coll Cardiol Intv 2015;8:245–53.
3. Dangas GD, George JC, Weintraub W. Popma JJ
Timing of staged percutaneous coronary inter-
vention in multivessel coronary artery disease.
J Am Coll Cardiol Intv 2010;3:1096–9.
4. Russo JJ, Wells GA, Chong AY, et al. Safety and
efficacy of staged percutaneous coronary inter-
vention during index admission for ST-elevation
myocardial infarction with multivessel coronary
disease (insights from the University of Ottawa
Heart Institute STEMI Registry). Am J Cardiol
2015;116:1157–62.
5. Kornowski R, Mehran R, Dangas G, et al., for the
HORIZONS-AMI Trial Investigators. Prognostic
impact of staged versus “one-time” multivessel
percutaneous intervention in acute myocardial
infarction: analysis from the HORIZONS-AMI
(Harmonizing Outcomes With Revascularization
and Stents in Acute Myocardial Infarction) trial.
J Am Coll Cardiol 2011;58:704–11.
6. Iqbal MB, Nadra IJ, Ding L, et al. Culprit vessel
versus multivessel versus in-hospital staged
intervention for patients with ST-segment eleva-
tion myocardial infarction and multivessel disease:
stratified analyses in high-risk patient groups and
anatomic subsets of nonculprit disease. J Am Coll
Cardiol Intv 2017;10:11–23.
7. Manari A, Varani E, Guastaroba P, et al. Long-
term outcome in patients with ST segment
elevation myocardial infarction and multivessel
disease treated with culprit-only, immediate, or
staged multivessel percutaneous revascularization
strategies: insights from the REAL registry. Cath-
eter Cardiovasc Interv 2014;84:912–22.
8. Levine GN, Bates ER, Blankenship JC, et al. 2015
ACC/AHA/SCAI focused update on primary percu-
taneous coronary intervention for patients with ST-
elevation myocardial infarction: an update of the
2011 ACCF/AHA/SCAI guideline for percutaneous
coronary intervention and the 2013 ACCF/AHA
OTMVR. We found that the decision to pursue staged PCI
was driven by specific patient characteristics, procedural
characteristics, and unmeasured site variation, and staged
PCI was associated with a significant mortality benefit.
WHAT IS NEXT? A randomized controlled trial evalu-
ating the efficacy and optimal timing of staged PCI is
needed to guide clinical practice.
guideline for the management of ST-elevation
myocardial infarction: a report of the American
College of Cardiology/American Heart Association
Task Force on Clinical Practice Guidelines and the
Society for Cardiovascular Angiography and In-
terventions. Circulation 2016;133:1135–47.
9. Box TL, McDonell M, Helfrich CD, et al. Stra-
tegies from a nationwide health information
technology implementation: the VA CART story.
J Gen Intern Med 2010;25 Suppl 1:72–6.
10. Byrd JB, Vigen R, Plomondon ME, et al. Data
quality of an electronic health record tool to
support VA cardiac catheterization laboratory
quality improvement: the VA Clinical Assessment,
Reporting, and Tracking System for Cath Labs
(CART) Program. Am Heart J 2013;165:434–40.
11. Brindis RG, Fitzgerald S, Anderson HV, Shaw RE,
Weintraub WS, Williams JF. The American College
of Cardiology-National Cardiovascular Data Regis-
try (ACC-NCDR): building a national clinical data
repository. J Am Coll Cardiol 2001;37:2240–5.
12. Glorioso TJ, Grunwald GK, Ho PM, Maddox TM.
Reference effect measures for quantifying,
comparing and visualizing variation from random and
fixed effects in non-normal multilevel models, with
applications to site variation in medical procedure use
and outcomes. BMC Med Res Methodol 2018;18:74.
13. Zhu Y, Ghosh D, Mitra N, Mukherjee B. A data-
adaptive strategy for inverse weighted estimation
of causal effects. Health Serv Outcomes Res
Methodol 2014;14:69–91.
14. Li F, Zaslavsky AM, Landrum MB. Propensity
score weighting with multilevel data. Stat Med
2013;32:3373–87.
15. Lin DY, Psaty BM, Kronmal RA. Assessing the
sensitivity of regression results to unmeasured
confounders in observational studies. Biometrics
1998;54:948–63.
16. Wald DS, Morris JK, Wald NJ, et al. Random-
ized trial of preventive angioplasty in myocardial
infarction. N Engl J Med 2013;369:1115–23.
17. Gershlick AH, Khan JN, Kelly DJ, et al. Ran-
domized trial of complete versus lesion-only
revascularization in patients undergoing primary
percutaneous coronary intervention for STEMI and
multivessel disease: the CvLPRIT trial. J Am Coll
Cardiol 2015;65:963–72.
Hu et al. 2273
Staged Versus One-Time Multivessel Revascularization
18. Engstrøm T, Kelbæk H, Helqvist S, et al. Com-
plete revascularisation versus treatment of the
culprit lesion only in patients with ST-segment
elevation myocardial infarction and multivessel dis-
ease (DANAMI-3-PRIMULTI): an open-label, rando-
mised controlled trial. Lancet 2015;386:665–71.
19. Smits PC, Boxma-de Klerk BM. Fractional flow
reserve-guided multivessel angioplasty in myocar-
dial infarction. N Engl J Med 2017;377:397–8.
20. Thiele H, Akin I, Sandri M, et al. PCI strategies in
patients with acute myocardial infarction and
cardiogenic shock. N Engl J Med 2017;377:2419–32.
21. Li Z, Zhou Y, Xu Q, Chen X. Staged versus one-
time complete revascularization with percuta-
neous coronary intervention in STEMI patients
with multivessel disease: a systematic review and
meta-analysis. PLoS ONE 2017;12:e0169406.
22. Vlaar PJ, Mahmoud KD, Holmes DR Jr., et al.
Culprit vessel only versus multivessel and staged
percutaneous coronary intervention for multi-
vessel disease in patients presenting with
ST-segment elevation myocardial infarction: a
pairwise and network meta-analysis. J Am Coll
Cardiol 2011;58:692–703.
23. Giacoppo D, Madhavan MV, Baber U, et al.
Impact of contrast-induced acute kidney injury
after percutaneous coronary intervention on
short- and long-term outcomes: pooled anal-
ysis from the HORIZONS-AMI and ACUITY trials.
Circ Cardiovasc Interv 2015;8:e002475.
24. Kong JA, Chou ET, Minutello RM, Wong SC,
Hong MK. Safety of single versus multi-vessel
angioplasty for patients with acute myocardial
infarction and multi-vessel coronary artery dis-
ease: report from the New York State Angioplasty
Registry. Coron Artery Dis 2006;17:71–5.
KEY WORDS multivessel coronary artery
disease, multivessel coronary
revascularization, one-time multivessel
revascularization, percutaneous coronary
intervention, staged percutaneous coronary
intervention
A PP END IX For supplemental tables and a
figure, please see the online version of this
paper.