ORIGINAL RESEARCH ARTICLE

ORIGINAL RESEARCH
Comparison Between Patent Ductus

ARTICLE
Arteriosus Stent and Modified Blalock-
Taussig Shunt as Palliation for Infants With
Ductal-Dependent Pulmonary Blood Flow
Insights From the Congenital Catheterization Research
Collaborative
Editorial, see p 602 Andrew C. Glatz, MD,
MSCE
BACKGROUND: Infants with ductal-dependent pulmonary blood flow may Christopher J. Petit, MD
undergo palliation with either a patent ductus arteriosus (PDA) stent or a Bryan H. Goldstein, MD
modified Blalock-Taussig (BT) shunt. A balanced multicenter comparison of Michael S. Kelleman, MS,
these 2 approaches is lacking. MSPH
Courtney E. McCracken,
METHODS: Infants with ductal-dependent pulmonary blood flow palliated PhD
with either a PDA stent or a BT shunt from January 2008 to November Alicia McDonnell, MD
2015 were reviewed from the 4 member centers of the Congenital Timothy Buckey, BS
Catheterization Research Collaborative. Outcomes were compared by use Christopher E. Mascio, MD
of propensity score adjustment to account for baseline differences between Subi Shashidharan, MD
groups. R. Allen Ligon, MD
Jingning Ao, BS
Downloaded from http://ahajournals.org by on October 2, 2018

RESULTS: One hundred six patients with a PDA stent and 251 patients Wendy Whiteside, MD
with a BT shunt were included. The groups differed in underlying anatomy W. Jack Wallen, MD
(expected 2-ventricle circulation in 60% of PDA stents versus 45% of Christina M. Metcalf, BA
BT shunts; P=0.001) and presence of antegrade pulmonary blood flow Varun Aggarwal, MD
(61% of PDA stents versus 38% of BT shunts; P<0.001). After propensity Hitesh Agrawal, MD
score adjustment, there was no difference in the hazard of the primary Athar M. Qureshi, MD
composite outcome of death or unplanned reintervention to treat cyanosis
(hazard ratio, 0.8; 95% confidence interval [CI], 0.52–1.23; P=0.31). Other
reinterventions were more common in the PDA stent group (hazard ratio,
29.8; 95% CI, 9.8–91.1; P<0.001). However, the PDA stent group had a
lower adjusted intensive care unit length of stay (5.3 days [95% CI, 4.2–6.7]
versus 9.19 days [95% CI, 7.9–10.6]; P<0.001), a lower risk of diuretic use
at discharge (odds ratio, 0.4; 95% CI, 0.25–0.64; P<0.001) and procedural
complications (odds ratio, 0.4; 95% CI, 0.2–0.77; P=0.006), and larger
Correspondence to: Andrew
(152 mm2/m2 [95% CI, 132–176] versus 125 mm2/m2 [95% CI, 113–138]; C. Glatz, MD, MSCE, Children’s
P=0.029) and more symmetrical (symmetry index, 0.84 [95% CI, 0.8–0.89] Hospital of Philadelphia, 34th St
versus 0.77 [95% CI, 0.75–0.8]; P=0.008] pulmonary arteries at the time of and Civic Center Blvd, 6th Floor,
Main Bldg, Philadelphia, PA 19104.
subsequent surgical repair or last follow-up. E-mail glatz@email.chop.edu
CONCLUSIONS: In this multicenter comparison of palliative PDA stent and BT Sources of Funding, see page 600
shunt for infants with ductal-dependent pulmonary blood flow adjusted for Key Words: Blalock-Taussig
differences in patient factors, there was no difference in the primary end point, procedure ◼ ductus arteriosus,
death or unplanned reintervention to treat cyanosis. However, other markers of patent ◼ heart diseases ◼ stents
morbidity and pulmonary artery size favored the PDA stent group, supporting ◼ propensity score

PDA stent as a reasonable alternative to BT shunt in select patients. © 2017 American Heart
Association, Inc.

Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987 February 6, 2018 589

 Glatz et al
Clinical Perspective
What Is New?
• Using a multicenter collaborative, we compared
surgical Blalock-Taussig shunt and transcatheter
patent ductus arteriosus stent as initial palliation
for infants with ductal-dependent pulmonary
blood flow.
• Although the observed risk of the primary outcome
(death or unplanned reintervention to treat cyano-
sis) was higher in the Blalock-Taussig shunt group,
there was no significant difference between groups
after adjustment for patient-level factors.
• After adjustment for patient factors, other out-
comes favored the patent ductus arteriosus stent
group, including fewer procedural complications,
shorter intensive care unit length of stay, less fre-
quent need for diuretics, and larger and more sym-
metrical pulmonary arteries at last follow-up.
What Are the Clinical Implications?
• These findings support patent ductus arteriosus
stent placement as a preferable alternative pallia-
tive strategy to Blalock-Taussig shunt placement in
select patients with ductal-dependent pulmonary
blood flow, particularly in experienced centers
where this procedure can be performed safely and
effectively.
• Further research is needed to determine specific
Downloaded from http://ahajournals.org by on October 2, 2018
anatomic characteristics of patients who would be
most likely to benefit from patent ductus arteriosus
stent placement.
I
nfants with various forms of congenital heart disease
that include ductal-dependent pulmonary blood flow
often require an early palliative procedure to augment
or establish a reliable source of pulmonary blood flow
until the suitable time when a more definitive reparative
or palliative surgical procedure can be performed. His-
torically, this has been most commonly accomplished
with a modified Blalock-Taussig (BT) shunt, although
the potential for morbidity and mortality associated
with this procedure has been recognized and persists.1–3
Patent ductus arteriosus (PDA) stent implantation was
introduced in 1992 as an alternative method to provide
a stable source of pulmonary blood flow.4 Since that
original description, there have been numerous reports
of the safety and feasibility of this approach,5–15 which
has the potential advantages of avoiding cardiothoracic
surgery in the neonatal period and stimulating better
and more symmetrical pulmonary artery growth.16–18
However, there are also potential disadvantages of a
PDA stent, including unique procedural complications,
concerns about the durability of stent patency,19 and
technical hurdles at subsequent surgical procedures.20
590 February 6, 2018
Direct comparisons of the BT shunt and PDA stent
as a palliative procedure for infants with ductal-depen-
dent pulmonary blood flow have been limited to a small
number of single-center studies21–23 without either the
sample size or case mix to make a meaningful com-
parison of relevant outcomes. In nonrandomized ob-
servational studies like these, there is risk for substantial
bias resulting from confounding by indication in which
a factor (or factors) influences both the particular treat-
ment strategy to which a patient is assigned and the
outcome of interest. In an attempt to minimize this po-
tential source of bias, we sought to compare the PDA
stent and BT shunt as palliative options for infants with
ductal-dependent pulmonary blood flow using a multi-
center collaborative that would provide both the sam-
ple size and practice variability needed to compare out-
comes adjusted for patient-level differences that may
affect assignment to a particular treatment strategy.
METHODS
The data, analytical methods, and study materials will not be
made available to other researchers for purposes of reproduc-
ing the results or replicating the procedure.
A retrospective cohort study was performed that included all
infants with ductal-dependent pulmonary blood flow and con-
fluent pulmonary arteries palliated at <1 year of age with either
a BT shunt (from January 1, 2012) or PDA stent (from January
1, 2008) through November 1, 2015, at the 4 member cen-
ters of the Congenital Catheterization Research Collaborative.
The differing inclusion dates for each treatment strategy were
intentionally chosen to capture the entire PDA stent experience
at all 4 centers while maintaining a contemporary cohort of
patients receiving a BT shunt at an approximate ratio of 2:1 (BT
shunt to PDA stent) to optimize statistical power. Patients were
excluded if there was a source of pulmonary blood flow other
than the ductus arteriosus or native pulmonary outflow tract
(eg, major aortopulmonary collateral vessels), if the branch pul-
monary arteries were discontinuous, or if additional surgical or
catheter-based interventions were required at the time of initial
palliation other than interventions on the pulmonary outflow
tract. All available medical records were reviewed, including all
follow-up data through December 24, 2015. The Congenital
Catheterization Research Collaborative, previously described,24
comprises investigators from the Children’s Hospital of
Philadelphia, Cincinnati Children’s Hospital Medical Center,
Children’s Healthcare of Atlanta, and Texas Children’s Hospital.
Individual centers performed the detailed medical record review
of their own patients. Deidentified data were then collected
and managed in a common central database with Research
Electronic Data Capture25 tools hosted at Children’s Healthcare
of Atlanta, which serves as the data coordinating center for the
Congenital Catheterization Research Collaborative. This study
was approved by the Institutional Review Board at each partici-
pating center with a waiver of the need for informed consent.
Outcomes
The primary outcome was defined a priori as a composite
of death or unplanned reintervention to treat cyanosis. An
Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987

unplanned reintervention to treat cyanosis was defined as
an intervention designed to increase oxygen saturation per-
formed during an otherwise unplanned catheterization or
surgical procedure that was prompted by clinically concerning
cyanosis. Given the variability in what may be considered clini-
cally concerning cyanosis between patients and providers, no
objective standard for oxygen saturation was used. Instead, if
the medical record made clear that the intervention was per-
formed in an unplanned manner because of concern for cya-
nosis, then the outcome was established. Cases in which this
outcome was less clearly established were reviewed on regular
study conference calls to ensure consistency in the manner in
which this outcome was classified. An unplanned reinterven-
tion to treat cyanosis was combined with death as a composite
end point with the notion that such an event could represent
a patient at risk for impending sudden cardiovascular death.
Secondary outcomes included the individual components of
the primary outcome, procedural complications, intensive care
unit (ICU) length of stay (LOS), hospital LOS, diuretic use at
hospital discharge, other reinterventions (both planned and
unplanned to treat concerns other than cyanosis), and branch
pulmonary artery size at last follow-up. The at-risk period for
developing any of these outcomes began at the time of the ini-
tial palliative procedure and ended at the time of subsequent
definitive surgical repair. For patients with expected 2-ventricle
physiology, the subsequent definitive surgical repair typically
consisted of a complete anatomic repair. For patients with
expected single-ventricle physiology, the subsequent defini-
tive surgical repair typically consisted of a palliative superior
cavopulmonary connection. For patients who did not undergo
a definitive surgical repair, the at-risk period extended to the
Downloaded from http://ahajournals.org by on October 2, 2018
point of last follow-up or study end date.
Pulmonary artery size was assessed by all imaging modalities
available at a time closest to definitive surgical repair or last fol-
low-up. For patients with pulmonary artery size measurements
made by echocardiography, magnetic resonance imaging,
or computed tomography, the dimensions of the pulmonary
arteries were abstracted from the imaging reports. For patients
with angiographic imaging of the pulmonary arteries, the pri-
mary images were reread by an expert reviewer at each center
(A.C.G., C.J.P., B.H.G., A.M.Q.) with the vessel diameter mea-
sured in a systolic frame at the hilum, proximal to the origin of
the upper lobe vessels. For patients who had pulmonary artery
size assessed by >1 imaging modality, the following prioritiza-
tion schema was created to select which measurement was used
in the analysis: angiography>computed tomography>magnetic
resonance imaging>echocardiography. Pulmonary artery size
was quantified in 2 ways: the Nakata index,26 the summed
cross-sectional area (assuming a cylindrical vessel) of the right
and left pulmonary arteries indexed to the patient’s body sur-
face area, and a measure we called the symmetry index, a ratio
of diameters of the smaller pulmonary artery to the larger pul-
monary artery. The symmetry index is always ≤1, with values
closer to 1 reflecting more symmetrical vessel size.
Statistical Analysis
Statistical analyses were performed by the Congenital
Catheterization Research Collaborative’s data coordinating
center using SAS version 9.4 (SAS Institute Inc, Cary, NC),
and statistical significance was assessed at the 0.05 level.
Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987
Multicenter Comparison of PDA Stent and BT Shunt
Normality of continuous variables was assessed with histo-
ORIGINAL RESEARCH
gram, normal probability plots, and the Anderson-Darling
test for normality. Descriptive statistics are presented as
counts and percentages for categorical variables and median
ARTICLE
(25th–75th percentile) for continuous data with skewed dis-
tributions. Continuous data were compared between patients
treated with a PDA stent and those treated with a BT shunt
with the use of Wilcoxon rank-sum tests, and comparisons
between categorical variables were performed with χ2 tests or
Fisher exact test when expected cell counts were <5.
Because groups were not equal at baseline (Table 1), inverse
probability of treatment weighting with propensity scores was
used to control for potential confounders and baseline differ-
ences between groups. The propensity score was estimated
with a logistic regression model in which treatment assign-
ment (PDA stent versus BT shunt, with BT shunt as the refer-
ence) was regressed on 6 variables thought to be associated
with treatment assignment: center, expected ultimate physi-
ology, presence of antegrade pulmonary blood flow, under-
lying anatomic diagnosis, preintervention inotrope use, and
preintervention mechanical ventilation (the model formulation
for the propensity score is shown in Table I in the online-only
Data Supplement). These 6 variables were selected a priori to
account for expected differences at the time a particular treat-
ment strategy was selected in patient anatomy, physiology,
and hemodynamic stability. To stabilize the weights, inverted
propensity scores were truncated at the 1st and 99th percen-
tiles and were normalized. Normalization consisted of divid-
ing each individual propensity score by the mean propensity
score of its respective treatment assignment. The standardized
mean difference was used to quantify the relative imbalance
in a covariate between the 2 treatment groups. A standardized
mean difference >0.25 was considered a clinically significant
imbalance. All adjusted models included the main effect of
treatment and were weighted by stabilized propensity score to
achieve balance between treatment groups.
To compare the effect of treatment (PDA stent versus BT
shunt) on the dichotomous outcomes (use of diuretics at dis-
charge and procedural complication), logistic regression was
used, and odds ratios with 95% confidence intervals (CIs) are
presented with BT shunt as the reference group. For the con-
tinuous outcomes (ICU LOS, hospital LOS, pulmonary artery
symmetry index, and Nakata index), residual errors were
gauged for normality via histograms and quantile-quantile
plots. Failing to meet this assumption, continuous outcomes
were log-transformed and reassessed for differences between
interventions. Model-based mean estimates were back-
transformed via exponentiation for interpretation purposes.
Results are presented as means with associated 95% CIs.
Time-dependent outcomes (death, unplanned reinter-
vention to treat cyanosis, and other reinterventions) were
analyzed with survival analysis. Because most of the events
occurred early in the follow-up period, the hazard was not
proportional with time, and parametric survival models
accounting for early failure were used. Parametric probability
estimates for time-dependent outcomes were obtained with
PROC HAZARD (available for download with the SAS sys-
tem at https://www.lerner.ccf.org/qhs/software/hazard/). The
HAZARD procedure uses maximum likelihood to resolve risk
distribution of time to event in up to 3 phases of risk (early
decreasing or peaking hazard, constant hazard, and late
February 6, 2018 591
 Glatz et al

Table 1.  Differences in Patient Demographic, Clinical, increasing hazard). Smoothed survival curves were generated
and Anatomic Factors Based on Treatment Strategy with the HAZPRED procedure in SAS. Death and unplanned
reintervention to treat cyanosis were also combined as a pri-
PDA Stent BT Shunt P
Covariate n (n=106, 29.7%) (n=251, 70.3%) Value mary composite outcome. Effect of treatment on the hazard
of the event of interest was quantified with the use of hazard
Center, n (%) 357 <0.001
ratios with associated 95% CIs. Because death was consid-
 1 26 (24.5) 67 (26.7) ered a competing event for reintervention, competing-risk
 2 36 (34.0) 69 (27.5) analysis was performed to model the probability over time
 3 36 (34.0) 49 (19.5) for 2 mutually exclusive end points after initial intervention,
death and reintervention, with the remaining patients being
 4 8 (7.5) 66 (26.3)
alive without reintervention. Separate models were con-
Intervention before BT 357 35 (33.0) 18 (7.2) <0.001 structed for unplanned reintervention to treat cyanosis and
shunt/PDA stent, n (%)
other reinterventions. To confirm the validity of this model-
Anatomic diagnosis, 357 <0.001 ing approach, time-dependent outcomes were also analyzed
n (%)
with Cox proportional hazard models with very similar results
 PA/IVS 47 (44.3) 50 (20) (Table II in the online-only Data Supplement).
 VSD/PS 26 (24.5) 62 (24.7)
 VSD/PA 18 (17) 99 (39.4)
 Tricuspid atresia 5 (4.7) 39 (15.5)
RESULTS
with PA or PS The cohort consisted of 106 patients treated with PDA
 Isolated PS 10 (9.4) 1 (0.4) stents and 251 patients treated with BT shunts. Differ-
Expected 2-ventricle 357 64 (60.4) 112 (44.6) 0.001 ences in baseline demographic, anatomic, and clinical
physiology, n (%) characteristics between treatment strategies are sum-
Antegrade pulmonary 357 65 (61.3) 95 (37.8) <0.001 marized in Table  1. There were significant differences
blood flow, n (%) in the distribution of treatment strategy across centers,
Gestational age, wk 347 38.0 (36.5–39.0) 38.0 (37.0–39.0) 0.41 as well as a number of patient-level factors. A higher
Prematurity (<37 wk), 352 27 (26.0) 53 (21.4) 0.35 proportion of patients with pulmonary atresia with in-
n (%) tact ventricular septum and isolated pulmonary stenosis
Birth weight, kg 350 2.9 (2.5–3.5) 3.0 (2.5–3.3) 0.91 were selected for PDA stent placement compared with
other anatomic diagnosis categories. Similarly, a higher
Downloaded from http://ahajournals.org by on October 2, 2018

Genetic syndrome, 355 13 (12.3) 38 (15.3) 0.46

n (%) proportion of patients with an expected 2-ventricle cir-
Other comorbid 355 13 (12.4) 49 (19.6) 0.12 culation and with antegrade pulmonary blood flow at
medical conditions, the time of initial palliation were selected for PDA stent
n (%)
placement. Despite this, a higher proportion of patients
Invasive ventilation 357 29 (27.4) 78 (31.1) 0.48
receiving prostaglandin infusion at the time of initial pal-
before intervention,
n (%) liation were treated with a PDA stent. Patients in both
Inotrope use before 357 8 (7.5) 29 (11.6) 0.26
groups not on prostaglandin at the time of initial pallia-
intervention, n (%) tion had either a ductus arteriosus that remained pat-
Prostaglandin use 357 96 (90.6) 203 (80.9) 0.02 ent without need for prostaglandin or a closed ductus
before intervention, with antegrade pulmonary blood flow but unaccept-
n (%) able cyanosis. Thus, the entire cohort was considered to
Age at intervention, d 357 9 (5–15) 6 (4–15) 0.2 have ductal-dependent pulmonary blood flow. Finally, a
Weight at 357 3.2 (2.7–3.7) 3.1 (2.7–3.5) 0.6 higher proportion of patients who had an intervention
intervention, kg before the initial palliation (the majority of which were
Preintervention echocardiography, n (%) interventions on the right ventricular outflow tract) were
 Systemic ventricular 357 105 (99.1) 247 (98.4) 1.0 treated with PDA stent placement. There were no sig-
function normal/ nificant differences in other baseline demographic, ana-
mildly depressed
tomic, and clinical characteristics between the 2 groups.
 Systemic 347 104 (98.1) 231 (95.9) 0.36
Of the patients palliated with a PDA stent, drug-eluting
atrioventricular
valve function coronary stents were used in 13 patients and bare metal
normal/mild coronary stents in the remainder.
regurgitation

Values are reported as median (25th–75th percentiles) when appropriate.

BT indicates Blalock-Taussig; IVS, intact ventricular septum; PA, pulmonary
atresia; PDA, patent ductus arteriosus; PS, pulmonary stenosis; and VSD,
ventricular septal defect.
Unadjusted Comparisons
Comparisons of observed (unadjusted) differences in
outcomes between groups are summarized in Table 2.
The primary composite outcome (death or unplanned

592 February 6, 2018 Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987

 Multicenter Comparison of PDA Stent and BT Shunt

Table 2.  Observed Differences in Outcomes Based on Treatment Strategy

ORIGINAL RESEARCH
PDA Stent BT Shunt
Primary and Secondary Outcomes n (n=106, 29.7%) (n=251, 70.3%) P Value

ARTICLE
Death or unplanned reintervention to treat cyanosis, n (%) 357 18 (17) 74 (29.5) 0.014
Death, n (%) 357 7 (6.6) 26 (10.4) 0.26
Unplanned reintervention to treat cyanosis, n (%) 357 12 (11.3) 52 (20.7) 0.035
Other reinterventions, n (%) 357 38 (35.8) 4 (1.6) <0.001
Procedural complications, n (%) 357 14 (13.2) 54 (21.5) 0.07
 Bleeding, not access related 0 24
 Arrhythmia 4 14
 Cardiac arrest 0 11
 ECMO 0 6
 Thrombosis, not access related 1 4
 Stroke 0 3
 Vascular injury, access related 10 0
 Bacteremia 1 0
 Stent migration 1 0
 Wound infection requiring repeat surgery 0 5
 Vocal cord paralysis 0 3
 Acute kidney injury 0 2
 Early reoperation 0 2
 Need for diaphragm plication 0 1
 Severe ventricular dysfunction 0 1
 Lung collapse 0 1
Duration of ICU LOS, d 354 4 (2–11) 7 (4–16) <0.001
Downloaded from http://ahajournals.org by on October 2, 2018

Total duration of hospital LOS, d 354 10 (6–21) 13 (9–27) <0.001

Diuretic use at discharge, n (%) 351 55 (53.4) 182 (73.4) <0.001
Pulmonary artery symmetry index 312 0.88 (0.76–0.96) 0.83 (0.71–0.92) 0.015
Nakata index, mm2/m2 220 158 (115–214) 131 (81–200) 0.016
Other outcomes
 Total duration of ventilation, d 355 1 (0–2) 3 (1–5) <0.001
 Duration of inotrope use, d 354 0 (0–0) 2 (1–4) <0.001
 Need for ECMO during recovery, n (%) 356 2 (1.9) 14 (5.6) 0.16
 Antiplatelet/anticoagulation use at hospital discharge, n (%) 357 97 (92) 235 (94) 0.47
 Age at definitive surgical repair, d 286 190 (137–294) 165 (131–217) 0.047
 Time to definitive surgical repair, d 287 178 (125–276) 150 (118–205) 0.043
Definitive surgical repair, n (%) 349 <0.001
 Stage 2 palliation 37 (35.9) 130 (52.8)
 Complete anatomic repair 31 (30.1) 77 (31.3)
 Transplantation 0 3 (1.2)
 None needed 21 (20.4) 4 (1.6)
 Not yet occurred or death before repair 14 (13.6) 32 (13)
Pulmonary artery plasty performed at definitive surgical repair, n (%) 283 33 (45.8) 89 (42.2) 0.59
Subsequent pulmonary artery intervention, n (%) 283 12 (16.7) 41 (19.2) 0.64

Values are reported as median (25th–75th percentile) when appropriate.

BT indicates Blalock-Taussig; ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit; LOS, length of stay; PDA, patent ductus
arteriosus; and plasty, surgical arterioplasty.

reintervention to treat cyanosis) occurred more com- increased rate of unplanned reintervention to treat
monly among the BT shunt group (29.5% versus 17%; cyanosis (20.7% versus 11.3%; P=0.035). Mortality
P=0.014). This difference was largely the result of an was higher in the BT shunt group, but this difference

Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987 February 6, 2018 593

 Glatz et al

was not statistically significant (10.4% versus 6.6%; Table 3.  Reinterventions After Initial Palliative
P=0.26). Among secondary outcomes, other reinter- Procedure
ventions occurred more commonly in the PDA stent PDA Stent BT Shunt
group (35.8% versus 1.6%; P<0.001). In the PDA Type of Reintervention n (%)* n (%)*
stent group, the ICU LOS and total hospital LOS were Unplanned reinterventions to treat cyanosis n=12 n=52
significantly shorter (median, 4 and 10 versus 7 and  Balloon angioplasty of PDA stent or BT 3 (25) 14 (26.9)
13 days, respectively; both P<0.001), and diuretic use shunt
at hospital discharge was less common (53.4% versus  Restenting of PDA stent or BT shunt 5 (41.7) 20 (38.5)
73.4%; P<0.001). Procedural complications occurred  Balloon angioplasty of pulmonary artery 1 (8.3) 8 (15.4)
more commonly in the BTS group, although this ob-
 Pulmonary artery stent 1 (8.3) 2 (3.8)
served difference did not reach statistical significance
 Systemic artery stent 0 (0) 1 (1.9)
(21.5% versus 13.2%; P=0.07). At the time of subse-
quent surgical repair or last follow-up, the pulmonary  Surgical revision of PDA stent or BT shunt 6 (50) 7 (13.5)

artery size in the PDA stent group was greater (median  Surgical pulmonary artery plasty 2 (16.7) 4 (7.7)
Nakata index, 158 versus 131 mm2/m2; P=0.016) and  RVOT stent 1 (8.3) 1 (1.9)
more symmetrical (median symmetry index, 0.88 ver-  Balloon atrial septostomy 0 (0) 3 (5.8)
sus 0.83; P=0.015).  Drainage of hemopericardium/ 0 (0) 2 (3.8)
The PDA stent group was older and had a longer hemothorax
time interval between the initial palliation and the de-  Urgent early superior cavopulmonary 1 (8.3) 2 (3.8)
finitive surgical procedure. This definitive operation was connection
less commonly a superior cavopulmonary connection in  Other 1 (8.3) 1 (1.9)
the PDA stent group. There were no significant differ- Other reinterventions (planned reinterventions or unplanned
ences in the need for pulmonary arterioplasty at the reinterventions to treat issues other than cyanosis)
time of the definitive surgical repair, nor were there dif-  Type of reintervention n=38* n=4*
ferences in the rates of subsequent interventions on the  Balloon angioplasty of PDA stent or BT 32 (84) 2 (50)
pulmonary arteries after definitive surgical repair. shunt
 Re-stenting of PDA stent or BT shunt 6 (15.8) 0 (0)

Types of Reinterventions  Balloon angioplasty of pulmonary artery 1 (2.6) 2 (50)

Downloaded from http://ahajournals.org by on October 2, 2018

 Pulmonary artery stent 1 (2.6) 0 (0)

A description of the types of reinterventions performed,
for both the unplanned reinterventions to treat cyano-  Surgical revision of PDA stent or BT shunt 1 (2.6) 0 (0)

sis and the other reinterventions (either planned rein-
terventions or unplanned reinterventions to treat issues
other than cyanosis), is summarized in Table 3. Among
both groups, the unplanned reinterventions to treat
cyanosis most commonly involved balloon or stent an-
gioplasty of the PDA stent or BT shunt or surgical revi-
sion. Other reinterventions occurred almost exclusively
among the PDA stent group and most commonly in-
cluded redilation of the stent by balloon angioplasty or,
in the case of patients with antegrade pulmonary blood
flow, repeat pulmonary valvuloplasty.
Propensity Score
The propensity score achieved balance among the
treatment groups because the standardized mean dif-
ferences after weighting were <25% for all variables
used in the propensity score except anatomic diagnosis,
which was nearly balanced with a standardized mean
difference of 0.31. In addition, the average stabilized
inverse probability of treatment weighting score was
close to 1 (mean, 1.0; SD, 0.73), indicating stability in
our weighting. Additional information on the propen-
sity score can be found in Table III in the online-only
Data Supplement.
 Pulmonary valvuloplasty 7 (18.4) 1 (25)
 Balloon angioplasty of RVOT stent 1 (2.6) 0 (0)
 Balloon atrial septostomy 0 (0) 1 (25)
BT indicates Blalock-Taussig; PDA, patent ductus arteriosus; plasty, surgical
arterioplasty; and RVOT, right ventricular outflow tract.
*Patients may have had >1 type of intervention at first reintervention, so
column counts do not add to total.
Propensity Score–Adjusted Comparisons
Models assessing propensity score–adjusted differ-
ences between treatment groups in primary and sec-
ondary outcomes are summarized in Tables 4 through
6. Adjusted differences in time-dependent primary
outcomes are shown in the survival models in Table 4.
After propensity score adjustment, there was no sig-
nificant difference in the hazard of the primary end
point (hazard ratio, 0.8; 95% CI, 0.52–1.23; P=0.31)
or in either of the individual components of the com-
posite end point. However, the significant difference
in the hazard for other reinterventions persisted, fa-
voring the BT shunt group (hazard ratio, 29.8; 95%
CI, 9.8–91.1; P<0.001). The comparison of survivor
experiences between groups for the primary compos-
ite outcome of death or unplanned reintervention to
treat cyanosis is depicted in Figure  1. A competing-

594 February 6, 2018 Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987

 Multicenter Comparison of PDA Stent and BT Shunt

Table 4.  Adjusted Effect of Treatment Strategy on Time-Dependent Outcomes With Survival

ORIGINAL RESEARCH
Models
Unadjusted Adjusted*

ARTICLE
n HR (95% CI)† P Value HR (95% CI)† P Value
Death or unplanned reintervention to treat cyanosis 357 0.53 (0.32–0.89) 0.02 0.8 (0.52–1.23) 0.31
Death 357 0.64 (0.28–1.47) 0.29 1.07 (0.52–2.18) 0.86
Unplanned reintervention to treat cyanosis‡ 357 0.51 (0.27–0.95) 0.03 0.75 (0.45–1.26) 0.28
Other reinterventions‡ 357 27.5 (9.8–77.2) <0.001 29.8 (9.8–91.1) <0.001

CI indicates confidence interval; and HR, hazard ratio.

*Adjusted for center, expected ultimate physiology, antegrade pulmonary blood flow, anatomic diagnosis, preintervention inotrope
use, and preintervention ventilation.
†Reference group is Blalock-Taussig shunt.
‡With death treated as a competing event.

risk analysis of death and unplanned reintervention
to treat cyanosis stratified by treatment group is dis-
played in Figure 2 and that for death and other rein-
terventions is shown in Figure 3.
Adjusted differences in dichotomous outcomes are
summarized in models shown in Table 5. The adjusted
risks of procedural complications (odds ratio, 0.4; 95%
CI, 0.2–0.77; P=0.006) and being prescribed a diuretic
at hospital discharge (odds ratio, 0.4; 95% CI, 0.25–
0.64; P<0.001) were significantly lower in the PDA
stent group.
Adjusted differences in continuous outcomes are
summarized in models shown in Table 6. The adjusted
ICU LOS and hospital LOS were shorter in the PDA
Downloaded from http://ahajournals.org by on October 2, 2018
in pulmonary artery size and symmetry between the
2 treatment groups are displayed graphically in Fig-
ures 4 and 5.
As a sensitivity analysis to assess whether the use
of varying inclusion time periods for the 2 comparison
groups introduced bias, all adjusted comparisons were
repeated with the exclusion of patients receiving PDA
stents before January 2, 2012. These results are largely
unchanged and presented in Tables IV through VI in the
online-only Data Supplement.
DISCUSSION
We report comparisons of a variety of clinically mean-

stent group (adjusted mean ICU LOS, 5.3 versus 9.1
days, P<0.001; hospital LOS, 13.5 versus 16.5 days,
P=0.06), although the difference in hospital LOS did
not attain statistical significance. Similarly, the pulmo-
nary arteries were larger (as measured by the Nakata
index) and more symmetrical in the PDA stent group
after propensity score adjustment. The differences
ingful outcomes in infants with ductal-dependent
pulmonary blood flow who were initially palliated
with either a PDA stent or a BT shunt. After adjust-
ment for patient factors, we found that there was no
difference between treatment strategies in the haz-
ard of our primary outcome, a composite of death or
unplanned reintervention to treat cyanosis. However,

Figure 1. Unadjusted parametric

survival estimates demonstrat-
ing freedom from the composite
primary outcome of death or
unplanned reintervention to treat
cyanosis based on treatment
strategy.
Solid lines represent the smoothed
survival curves; dashed lines show
the 95% confidence intervals.
Orange and black lines represent es-
timates of reintervention-free survival
for patients receiving patent ductus
arteriosus (PDA) stents and Blalock-
Taussig (BT) shunts, respectively.

Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987 February 6, 2018 595

 Glatz et al
those treated with a PDA stent had a lower risk of
procedural complications, shorter ICU LOS, less use
of diuretics, and larger and more symmetrical pulmo-
nary arteries before their subsequent surgical repair
or palliation. Notably, other reinterventions (both
planned and unplanned to treat concerns other than
cyanosis) occurred more commonly in the PDA stent
group. This is by far the largest reported cohort of in-
fants treated with PDA stent and the first from a mul-
ticenter collaborative. The overall size of the cohort
and the variability in practice patterns across centers
allowed comparison of the 2 treatment strategies ad-
justed with propensity score methods to offset the
potential bias related to confounding by indication.
In the absence of a prospective study in which pa-
tients are randomly assigned to treatment strategy,
Downloaded from http://ahajournals.org by on October 2, 2018
we believe that this is the best possible method to
compare these 2 treatment strategies in a relatively
balanced fashion.
We chose a priori a composite of death or un-
planned reintervention to treat cyanosis as the primary
outcome, with the notion that cyanosis requiring rein-
tervention is a potential harbinger of a life-threatening
event such as acute thrombotic shunt or stent occlu-
sion. The observed risk of this composite outcome was
17% in the PDA stent group and 29.5% in the BT
shunt group, a difference that was not statistically dif-
ferent after propensity score adjustment. It is impor-
tant to interpret this finding in the context that the
results of the PDA stent group are derived from a time
596 February 6, 2018
Figure 2. Competing-risk analysis
for death or unplanned reinter-
vention to treat cyanosis, strati-
fied by treatment strategy.
Solid lines represent unplanned
reintervention for cyanosis; dashed
lines represent the competing event,
death. Orange and black lines rep-
resent event estimates for patients
receiving patent ductus arteriosus
(PDA) stents and Blalock-Taussig (BT)
shunts, respectively.
period that covers the entire PDA stent placement ex-
perience at all centers, and thus, any “learning curve”
is included in these results. Examining the individual
components of this composite outcome shows that
the observed risk of death was 6.6% in the PDA stent
group and 10.4% in the BT shunt group, again a dif-
ference that was not statistically significant. It is impor-
tant to note that this measure incorporates all-cause
mortality, not only those deaths resulting from BT
shunt or PDA stent dysfunction. These mortality risks
are comparable to what has been previously reported
and underscore the fact that patients with shunt- or
ductal stent–dependent pulmonary blood flow remain
a vulnerable population. In a large multicenter report
from the Society of Thoracic Surgeons Congenital
Heart Surgery Database, a cohort of neonates who
underwent BT shunt placement without concomi-
tant procedures experienced in-hospital mortality at a
rate of 7.2% with a 13.1% rate of serious morbid-
ity.2 Although the at-risk period for patients in that
cohort was shorter than what is included in the pres-
ent study, both reports suggest that shunt procedures
to provide pulmonary blood flow are associated with
a high risk of early postoperative mortality and mor-
bidity, even when compared with other types of neo-
natal cardiac surgery. Because this Society of Thoracic
Surgeons Congenital Heart Surgery cohort excluded
neonates who received other procedures (besides PDA
ligation), it is likely these were predominantly patients
with ductal-dependent pulmonary blood flow and
Figure 3. Competing-risk analysis
for death or other reinterven-
tions, stratified by treatment
strategy.
Solid lines represent other reinter-
ventions; dashed lines represent the
competing event, death. Orange and
black lines represent event estimates
for patients receiving patent ductus
arteriosus (PDA) stents and Blalock-
Taussig (BT) shunts, respectively.
Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987
 Multicenter Comparison of PDA Stent and BT Shunt

Table 5.  Adjusted Effect of Treatment Strategy on standard practice that has developed at some of the

ORIGINAL RESEARCH
Dichotomous Outcomes participating centers whereby elective cardiac catheter-
Unadjusted Adjusted* ization is performed routinely at defined intervals after

n OR (95% CI)† P Value
Procedural 357 0.56 0.07
complications (0.29–1.05)
Diuretic use 351 0.42 <0.001
at discharge (0.26–0.67)
CI indicates confidence interval; and OR, odds ratio.
*Adjusted for center, expected ultimate physiology, antegrade pulmonary
blood flow, anatomic diagnosis, preintervention inotrope use, and preintervention
ventilation.
†Reference group is Blalock-Taussig shunt.
thus quite comparable to the cohort in the present
study. Of note, 2 of the main risk factors identified for
mortality in that report were preoperative mechani-
cal ventilation and underlying cardiac anatomy, factors
that were included in our propensity-adjusted models.
In an older study restricted to patients with tetralogy
of Fallot, mortality after BT shunt was 8.3%.27 Finally,
using an audit data set from the United Kingdom to
capture a large cohort of patients who had surgery
for placement of an isolated BT shunt, Dorobantu et
al3 reported an overall mortality rate of 13.9% and
a 17.8% risk of shunt reintervention. Similar rates of
unexpected shunt reintervention have also been re-
ported in large single-center cohorts.1 Data on mortal-
Downloaded from http://ahajournals.org by on October 2, 2018
ARTICLE
OR (95%
initial PDA stent implantation. At such procedures, bal-
CI)† P Value loon angioplasty of or additional stent implantation in
0.4 0.006 the existing PDA stent may occur to optimize pulmo-
(0.2–0.77) nary blood flow, even though the patient may not be
0.4 <0.001 more cyanotic than previously. It is possible that such
(0.25–0.64) planned reinterventions could affect the rate at which
these patients would have developed worsening cya-
nosis and thus needed an unplanned reintervention to
treat cyanosis. This practice is driven partially by the
observation that PDA stents have a tendency to devel-
op neointimal proliferation over time, a phenomenon
that may affect their patency and durability, as high-
lighted by Sivakumar et al.19 In that report, patients
with 2-ventricle anatomies initially palliated with a PDA
stent required complete surgical repair earlier than
the typical practice at that center because of in-stent
stenosis and inadequate pulmonary blood flow. This
may be an issue intrinsic to stent placement in ductal
tissue because there was no difference in the use of
antiplatelet or anticoagulation regimens between the
groups in the present study that could explain the dif-
fering reintervention rates. Despite this concern, the
duration from palliation to definitive surgical repair in
the present study was significantly longer in the PDA
stent group, suggesting that a PDA stent can be as du-
rable as BT shunt placement, recognizing that elective

ity and reintervention rates after PDA stent placement

are much scarcer because this procedure is newer and
performed less commonly. In a handful of smaller sin-
gle-center studies, reported mortality rates after PDA
stent have ranged from 3.6% to 31%.5,6,8,10,12,13,15
We chose to separate reinterventions into those
that were unplanned to treat cyanosis versus reinter-
ventions for other purposes (either planned reinterven-
tions or unplanned reinterventions for issues other than
cyanosis), so direct comparisons with past publications
are difficult. Although the rate of unplanned reinter-
ventions to treat cyanosis was not different between
the 2 groups, it is worth noting that the rate of other
reinterventions was significantly higher among the
PDA stent group. This results, at least in part, from the
PDA stent reintervention may be used in the interim.
Redilation of the PDA stents was not the only type of
reintervention performed, however. Other reinterven-
tions included those needed for recurrent right ven-
tricular outflow tract obstruction in patients who had
previously had an outflow tract intervention. Although
these reinterventions were not deemed to be a harbin-
ger of potential sudden cardiac death in this analysis,
these procedures represent a real cost to the patient,
in both clinical and economic terms, and thus need to
be acknowledged.
Other secondary outcomes, including ICU LOS and
use of diuretic at hospital discharge, all favored the
PDA stent group, even after adjustment for patient fac-

Table 6.  Adjusted Effect of Treatment Strategy on Continuous Outcomes

Unadjusted Adjusted*
n PDA Stent BT Shunt P Value PDA Stent BT Shunt P Value
ICU LOS, d 354 4.4 (3.5–5.6) 8.9 (7.7–10.4) <0.001 5.3 (4.2–6.7) 9.1 (7.9–10.6) <0.001
Hospital LOS, d 354 11.6 (9.8–13.9) 16.2 (14.5–18.2) 0.002 13.5 (11.3–16.1) 16.5 (14.8–18.5) 0.06
Pulmonary artery symmetry index 312 0.83 (0.79–0.88) 0.77 (0.75–0.80) 0.02 0.84 (0.8–0.89) 0.77 (0.75–0.8) 0.008
Nakata index, mm2/m2 220 161 (138–187) 130 (118–143) 0.02 152 (132–176) 125 (113–138) 0.029

Values are displayed as mean (95% confidence interval).

BT indicates Blalock-Taussig; ICU, intensive care unit; LOS, length of stay; and PDA, patent ductus arteriosus.
*Adjusted for center, expected ultimate physiology, antegrade pulmonary blood flow, anatomic diagnosis, preintervention inotrope use, and preintervention
ventilation.

Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987 February 6, 2018 597

 Glatz et al
Figure 4. Box-and-whiskers plot demonstrating the
difference in pulmonary artery size as measured by the
Nakata index at the time of definitive surgical repair or
last follow-up between the 2 treatment groups.
The center line of the box represents the unadjusted median;
edges of the box, the unadjusted 25th and 75th percentile
values; whiskers, the unadjusted 1.5×interquartile range
values; and circles, the outliers. The diamond represents the
propensity-adjusted mean value. For ease of visual display,
values >95th percentile were truncated. The P value is the
Downloaded from http://ahajournals.org by on October 2, 2018
difference in the adjusted model. BT indicates Blalock-Tauss-
ig; and PDA, patent ductus arteriosus.
tors. The increased risk of being discharged on diuret-
ics in the BT shunt group suggests that these patients
were perceived to have excessive pulmonary blood or
residual pulmonary edema. Alternatively, this differ-
ence could reflect differences in practice patterns in
the management of postoperative patients compared
with postcatheterization patients. Other measures of
hospital-related morbidity such as duration of mechan-
ical ventilation and inotrope use also favored the PDA
stent. These findings are not entirely unanticipated
given this comparison between invasive and minimally
invasive treatment strategies. There have been a few
previous head-to-head comparisons of patients receiv-
ing a PDA stent and those receiving a surgical shunt,
which are worth examining in more detail. In 2012,
Amoozgar et al21 compared the short-term outcomes
of 18 infants with ductal-dependent pulmonary blood
flow palliated with a PDA stent at 1 institution with
20 infants treated with surgical systemic-pulmonary
shunts at a second institution and found no differences
between the 2 groups. McMullan et al23 found a similar
rate of reintervention between 13 infants with a PDA
stent and 42 with a BT shunt at a single center but
a higher rate of procedural complications and branch
598 February 6, 2018
Figure 5. Box-and-whiskers plot demonstrating the dif-
ference in pulmonary artery symmetry as measured by
the symmetry index (ratio of smaller pulmonary artery
to larger pulmonary artery) at the time of definitive
surgical repair or last follow-up between the 2 treat-
ment groups.
The center line of the box represents the unadjusted median;
edges of the box, the unadjusted 25th and 75th percentile
values; whiskers, the unadjusted 1.5×interquartile range
values; and circles, the outliers. The diamond represents the
propensity-adjusted mean value. The P value is the differ-
ence in the adjusted model. BT indicates Blalock-Taussig; and
PDA, patent ductus arteriosus.
pulmonary artery stenosis in the BT shunt group. It is
worth noting that these groups differed in underlying
cardiac anatomy (primarily in the proportion of pa-
tients with pulmonary atresia). Finally, Mallula et al22
compared 13 patients with a PDA stent with 16 with
a BT shunt in a single center and found a shorter du-
ration of hospitalization in the PDA stent group with
equivalency in other short-term outcomes. However,
the BT shunt group was significantly smaller in weight
in this center. Comparisons adjusted for the differences
in patients between groups were not performed in any
of these studies.
With similarity in the primary outcome, the benefit
in other short-term outcomes may favor PDA stent as
the preferred strategy. However, short-term outcomes
should not be the only focus, particularly when mor-
tality is relatively uncommon. Longer-term outcomes
in infants with ductal-dependent pulmonary blood
flow are more challenging to measure but are likely
to be influenced by the overall growth of the pulmo-
nary arteries and pulmonary vascular bed. In both
single-ventricle28,29 and 2-ventricle30 anatomies, better
growth and size of the pulmonary arteries have been
associated with improved longer-term outcomes. In
Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987

the present study, we found that infants palliated with
PDA stent had larger (as measured by Nakata index)
and more symmetrical pulmonary arteries at the time
of subsequent surgical repair, a difference that persist-
ed after adjustment for patient factors. This issue has
been previously explored in depth by Santoro and col-
leagues.9–11,16–18 Over a series of articles, they demon-
strated a significant symmetrical increase in pulmonary
artery size in a cohort of patients with a PDA stent,10,18
a consistent finding in both low- and normal-weight
infants9 and in infants with severely hypoplastic pulmo-
nary arteries17 at the time of initial palliation. Further-
more, although the overall pulmonary artery growth
in a contemporary cohort of infants palliated with BT
shunts was comparable, growth was more symmetrical
in the PDA stent group.16 This difference in pulmonary
artery growth may reflect differences in pulmonary ar-
tery blood flow characteristics. In our PDA stent cohort,
the presence of antegrade pulmonary blood flow was
more common and may have enhanced pulmonary
artery size, although this factor was included in the
propensity score adjustment. In addition, blood flow
through the PDA stent enters the pulmonary arteries
centrally, with the potential for relatively symmetrical
flow to each branch pulmonary artery. In the case of
a BT shunt, flow typically enters the pulmonary artery
bed asymmetrically, at the site of surgical shunt inser-
tion into one of the branch pulmonary arteries, with
the risk of unilateral pulmonary artery stenosis at the
Downloaded from http://ahajournals.org by on October 2, 2018
site of shunt insertion further exacerbating the poten-
tial for flow asymmetry.
An additional marker of long-term pulmonary ar-
tery size and function can be measured in the need
for additional interventions on the pulmonary arteries
(either surgical or transcatheter). There may be a num-
ber of possible mechanisms behind the development
of pulmonary artery stenosis requiring later reinterven-
tion, including the presence of shunt or stent mate-
rial in the pulmonary artery or constriction of ductal
tissue after stent or shunt placement that results in
unilateral branch pulmonary artery stenosis. This issue,
as it relates to PDA stent placement, was highlighted
by Vida et al20 in a cohort of 13 patients treated with
PDA stent, of whom nearly half required additional
pulmonary arterioplasty at the time of complete sur-
gical repair and 4 needed additional later pulmonary
artery interventions. In the present study, although the
use of pulmonary arterioplasty at subsequent surgi-
cal repair exceeded 40%, there was no difference in
this rate between treatment groups. Similarly, the rate
of subsequent interventions on the pulmonary arter-
ies was not different between the PDA stent and BT
shunt groups.
The risk of procedural complications was significant-
ly higher in the BT shunt group after adjustment for
patient factors. In addition, the complications observed
Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987
Multicenter Comparison of PDA Stent and BT Shunt
in the BT shunt group tended to be graver, including
ORIGINAL RESEARCH
the need for extracorporeal membrane oxygenation,
cardiac arrest, stroke, and wound infection requiring
ARTICLE
repeat surgery. In contrast, the vast majority of com-
plications related to PDA stent procedures were vascu-
lar, most commonly occlusion of the femoral artery at
the site of cannulation. This is a complication unique
to the PDA stent group and is readily manageable with
resolution of arterial occlusion in most patients.31 The
approach to PDA stent implantation may also require
alternative access strategies, including the use of the
common carotid artery or axillary artery for the more
vertically oriented PDA.32,33
A number of limitations to the present study deserve
mention. Although we attempted to adjust for all poten-
tially relevant confounding variables, the retrospective
nature of the study allows the possibility of unmeasured
confounders. Specifically, we are unable to account for
differences in postintervention management that may
have affected outcomes. The PDA stent group includes
only subjects who had technically successful stent place-
ment. We were unable to identify patients who may
have had an unsuccessful attempt at PDA stent place-
ment or were not even considered for PDA stent on the
basis of noninvasive prepalliation imaging. It is certainly
possible that certain ductal anatomies may not be appro-
priate for PDA stent implantation, although it is beyond
the scope and design of this outcomes study to explore
such important technical/feasibility issues in more detail.
Finally, reliable measures of pulmonary artery size at birth
were not available, so we are not able to comment spe-
cifically on pulmonary artery “growth” but rather on the
size of the pulmonary arteries at the time of definitive
surgical repair or last follow-up. It is worth noting, how-
ever, that pulmonary artery size and symmetry continued
to favor the PDA stent group after adjustment for factors
(expected ultimate physiology, anatomic diagnosis, and
the presence of antegrade pulmonary blood flow) that
might be expected to at least partially account for base-
line differences in pulmonary artery size.
CONCLUSIONS
In this large multicenter comparison of a BT shunt and
a PDA stent as palliation for infants with ductal-depen-
dent pulmonary blood flow, the first of its kind, we
found no difference in the adjusted hazard of death or
unplanned reintervention to treat cyanosis. Other rein-
terventions occurred more commonly in the PDA stent
group. However, in propensity score–adjusted analysis,
the risk of procedural complications and diuretic use at
hospital discharge were lower, ICU LOS was shorter, and
pulmonary artery size was larger and more symmetrical
in the PDA stent group. In sum, we believe that these
findings support the use of PDA stent as a reasonable
February 6, 2018 599
 Glatz et al
alternative palliative therapy to BT shunt in selected pa-
tients with ductal-dependent pulmonary blood flow.
SOURCES OF FUNDING
None.
DISCLOSURES
None.
AFFILIATIONS
Cardiac Center at the Children’s Hospital of Philadelphia,
University of Pennsylvania School of Medicine (A.C.G.,
A.M., T.B., C.E.M.). Children’s Healthcare of Atlanta, Emory
University School of Medicine, GA (C.J.P., M.S.K., C.E.M.,
S.S., R.A.L., J.A.). Heart Institute, Cincinnati Children’s
Hospital Medical Center, University of Cincinnati College of
Medicine, OH (B.H.G., W.W., W.J.W., C.M.M.). Lillie Frank
Abercrombie Section of Cardiology, Texas Children’s Hospital,
Baylor College of Medicine, Houston (V.A., H.A., A.M.Q.).
FOOTNOTES
Received June 16, 2017; accepted October 10, 2017.
The online-only Data Supplement is available with this ar-
ticle at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/
Downloaded from http://ahajournals.org by on October 2, 2018
CIRCULATIONAHA.117.029987/-/DC1.
Circulation is available at http://circ.ahajournals.org.
REFERENCES
1. O’Connor MJ, Ravishankar C, Ballweg JA, Gillespie MJ, Gaynor JW, Tab-
butt S, Dominguez TE. Early systemic-to-pulmonary artery shunt interven-
tion in neonates with congenital heart disease. J Thorac Cardiovasc Surg.
2011;142:106–112. doi: 10.1016/j.jtcvs.2010.10.033.
2. Petrucci O, O’Brien SM, Jacobs ML, Jacobs JP, Manning PB, Eghtesady P.
Risk factors for mortality and morbidity after the neonatal Blalock-Taussig
shunt procedure. Ann Thorac Surg. 2011;92:642–651. doi: 10.1016/j.
athoracsur.2011.02.030.
3. Dorobantu DM, Pandey R, Sharabiani MT, Mahani AS, Angelini GD, Mar-
tin RP, Stoica SC. Indications and results of systemic to pulmonary shunts:
results from a national database. Eur J Cardiothorac Surg. 2016;49:1553–
1563. doi: 10.1093/ejcts/ezv435.
4. Gibbs JL, Rothman MT, Rees MR, Parsons JM, Blackburn ME, Ruiz CE.
Stenting of the arterial duct: a new approach to palliation for pulmonary
atresia. Br Heart J. 1992;67:240–245.
5. Alwi M, Choo KK, Latiff HA, Kandavello G, Samion H, Mulyadi MD. Initial
results and medium-term follow-up of stent implantation of patent duc-
tus arteriosus in duct-dependent pulmonary circulation. J Am Coll Cardiol.
2004;44:438–445. doi: 10.1016/j.jacc.2004.03.066.
6. Gewillig M, Boshoff DE, Dens J, Mertens L, Benson LN. Stenting the neo-
natal arterial duct in duct-dependent pulmonary circulation: new tech-
niques, better results. J Am Coll Cardiol. 2004;43:107–112.
7. Michel-Behnke I, Akintuerk H, Thul J, Bauer J, Hagel KJ, Schranz D. Stent
implantation in the ductus arteriosus for pulmonary blood supply in con-
genital heart disease. Catheter Cardiovasc Interv. 2004;61:242–252. doi:
10.1002/ccd.10766.
8. Odemis E, Haydin S, Guzeltas A, Ozyilmaz I, Bilici M, Bakir I. Stent implan-
tation in the arterial duct of the newborn with duct-dependent pulmonary
circulation: single centre experience from Turkey. Eur J Cardiothorac Surg.
2012;42:57–60. doi: 10.1093/ejcts/ezr258.
600 February 6, 2018
9. Santoro G, Gaio G, Castaldi B, Palladino MT, Iacono C, Russo MG, Calabrò
R. Arterial duct stenting in low-weight newborns with duct-dependent
pulmonary circulation. Catheter Cardiovasc Interv. 2011;78:677–685. doi:
10.1002/ccd.23076.
10. Santoro G, Gaio G, Giugno L, Capogrosso C, Palladino MT, Iacono C,
Caianiello G, Russo MG. Ten-years, single-center experience with arte-
rial duct stenting in duct-dependent pulmonary circulation: early results,
learning-curve changes, and mid-term outcome. Catheter Cardiovasc In-
terv. 2015;86:249–257. doi: 10.1002/ccd.25949.
11. Santoro G, Gaio G, Palladino MT, Iacono C, Carrozza M, Esposito R, Russo
MG, Caianiello G, Calabrò R. Stenting of the arterial duct in newborns
with duct-dependent pulmonary circulation. Heart. 2008;94:925–929.
doi: 10.1136/hrt.2007.123000.
12. Schneider M, Zartner P, Sidiropoulos A, Konertz W, Hausdorf G. Stent
implantation of the arterial duct in newborns with duct-dependent circu-
lation. Eur Heart J. 1998;19:1401–1409.
13. Schranz D, Michel-Behnke I, Heyer R, Vogel M, Bauer J, Valeske K,
Akintürk H, Jux C. Stent implantation of the arterial duct in new-
borns with a truly duct-dependent pulmonary circulation: a single-
center experience with emphasis on aspects of the interventional
technique. J Interv Cardiol. 2010;23:581–588. doi: 10.1111/j.1540-
8183.2010.00576.x.
14. Udink Ten Cate FE, Sreeram N, Hamza H, Agha H, Rosenthal E, Qureshi
SA. Stenting the arterial duct in neonates and infants with congenital
heart disease and duct-dependent pulmonary blood flow: a multicenter
experience of an evolving therapy over 18 years. Catheter Cardiovasc In-
terv. 2013;82:E233–E243. doi: 10.1002/ccd.24878.
15. Celebi A, Yucel IK, Bulut MO, Kucuk M, Balli S. Stenting of the ductus
arteriosus in infants with functionally univentricular heart disease and
ductal-dependent pulmonary blood flow: a single-center experience.
Catheter Cardiovasc Interv. 2017;89:699–708. doi: 10.1002/ccd.26796.
16. Santoro G, Capozzi G, Caianiello G, Palladino MT, Marrone C, Farina G,
Russo MG, Calabrò R. Pulmonary artery growth after palliation of congen-
ital heart disease with duct-dependent pulmonary circulation: arterial duct
stenting versus surgical shunt. J Am Coll Cardiol. 2009;54:2180–2186.
doi: 10.1016/j.jacc.2009.07.043.
17. Santoro G, Gaio G, Capozzi G, Giugno L, Palladino MT, Capogrosso C,
D’Aiello AF, Caianiello G, Russo MG. Fate of hypoplastic pulmonary ar-
teries after arterial duct stenting in congenital heart disease with duct-
dependent pulmonary circulation. JACC Cardiovasc Interv. 2015;8:1626–
1632. doi: 10.1016/j.jcin.2015.05.027.
18. Santoro G, Palladino MT, Capozzi G, Iacono C, Russo MG, Calabrò R. Pul-
monary artery growth following arterial duct stenting in congenital heart
disease with duct-dependent pulmonary circulation. Catheter Cardiovasc
Interv. 2009;74:1072–1076. doi: 10.1002/ccd.22191.
19. Sivakumar K, Bhagyavathy A, Coelho R, Satish R, Krishnan P. Longevity of
neonatal ductal stenting for congenital heart diseases with duct-depen-
dent pulmonary circulation. Congenit Heart Dis. 2012;7:526–533. doi:
10.1111/j.1747-0803.2012.00657.x.
20. Vida VL, Speggiorin S, Maschietto N, Padalino MA, Tessari C, Biffanti R,
Cerutti A, Milanesi O, Stellin G. Cardiac operations after patent ductus
arteriosus stenting in duct-dependent pulmonary circulation. Ann Thorac
Surg. 2010;90:605–609. doi: 10.1016/j.athoracsur.2010.04.007.
21. Amoozgar H, Cheriki S, Borzoee M, Ajami G, Soltani M, Ahmadipour M,
Peiravian F, Amirghofran A. Short-term result of ductus arteriosus stent
implantation compared with surgically created shunts. Pediatr Cardiol.
2012;33:1288–1294. doi: 10.1007/s00246-012-0304-x.
22. Mallula K, Vaughn G, El-Said H, Lamberti JJ, Moore JW. Comparison of
ductal stenting versus surgical shunts for palliation of patients with pul-
monary atresia and intact ventricular septum. Catheter Cardiovasc Interv.
2015;85:1196–1202. doi: 10.1002/ccd.25870.
23. McMullan DM, Permut LC, Jones TK, Johnston TA, Rubio AE. Modified
Blalock-Taussig shunt versus ductal stenting for palliation of cardiac le-
sions with inadequate pulmonary blood flow. J Thorac Cardiovasc Surg.
2014;147:397–401. doi: 10.1016/j.jtcvs.2013.07.052.
24. Petit CJ, Glatz AC, Qureshi AM, Sachdeva R, Maskatia SA, Justino H,
Goldberg DJ, Mozumdar N, Whiteside W, Rogers LS, Nicholson GT, Mc-
Cracken C, Kelleman M, Goldstein BH. Outcomes after decompression of
the right ventricle in infants with pulmonary atresia with intact ventricular
septum are associated with degree of tricuspid regurgitation: results from
the Congenital Catheterization Research Collaborative. Circ Cardiovasc
Interv. 2017;10:e004428.
25. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research
Electronic Data Capture (REDCap): a metadata-driven methodology and
Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987

workflow process for providing translational research informatics support.
J Biomed Inform. 2009;42:377–381. doi: 10.1016/j.jbi.2008.08.010.
26. Nakata S, Imai Y, Takanashi Y, Kurosawa H, Tezuka K, Nakazawa M,
Ando M, Takao A. A new method for the quantitative standardization
of cross-sectional areas of the pulmonary arteries in congenital heart dis-
eases with decreased pulmonary blood flow. J Thorac Cardiovasc Surg.
1984;88:610–619.
27. Gladman G, McCrindle BW, Williams WG, Freedom RM, Benson LN. The
modified Blalock-Taussig shunt: clinical impact and morbidity in Fallot’s
tetralogy in the current era. J Thorac Cardiovasc Surg. 1997;114:25–30.
doi: 10.1016/S0022-5223(97)70113-2.
28. Knott-Craig CJ, Julsrud PR, Schaff HV, Puga FJ, Danielson GK. Pulmonary
artery size and clinical outcome after the modified Fontan operation. Ann
Thorac Surg. 1993;55:646–651.
29. Ovroutski S, Ewert P, Alexi-Meskishvili V, Hölscher K, Miera O, Peters B,
Hetzer R, Berger F. Absence of pulmonary artery growth after Fontan
Downloaded from http://ahajournals.org by on October 2, 2018
Circulation. 2018;137:589–601. DOI: 10.1161/CIRCULATIONAHA.117.029987
Multicenter Comparison of PDA Stent and BT Shunt
operation and its possible impact on late outcome. Ann Thorac Surg.
2009;87:826–831. doi: 10.1016/j.athoracsur.2008.10.075.
ORIGINAL RESEARCH
30. Sandoval JP, Chaturvedi RR, Benson L, Morgan G, Van Arsdell G, Honjo O,
Caldarone C, Lee KJ. Right ventricular outflow tract stenting in tetralogy
ARTICLE
of Fallot infants with risk factors for early primary repair. Circ Cardiovasc
Interv. 2016;9:e003979.
31. Glatz AC, Keashen R, Chang J, Balsama LA, Dori Y, Gillespie MJ, Giglia
TM, Raffini L, Rome JJ. Outcomes using a clinical practice pathway for
the management of pulse loss following pediatric cardiac catheterization.
Catheter Cardiovasc Interv. 2015;85:111–117. doi: 10.1002/ccd.25686.
32. Choudhry S, Balzer D, Murphy J, Nicolas R, Shahanavaz S. Percutaneous
carotid artery access in infants < 3 months of age. Catheter Cardiovasc
Interv. 2016;87:757–761. doi: 10.1002/ccd.26310.
33. Justino H, Petit CJ. Percutaneous common carotid artery access for pe-
diatric interventional cardiac catheterization. Circ Cardiovasc Interv.
2016;9:e003003. doi: 10.1161/CIRCINTERVENTIONS.115.003003.
February 6, 2018 601