2. same ultrasonography equipment, and recorded on video
for retrospective analysis. Measurements of the aortic arch
were obtained by two-dimensional echocardiography at the
end of systole from the suprasternal notch view, after
calibrating the system using the two-dimensional centime-
ter scale. Morphologic parameters and distances (d1 to d12)
were measured by three independent observers (S.O,
S.D.B, and F.B.) as described in Figure 1, and noted sepa-
rately. During the same time frame, 23 controls (16 neonates
and 7 infants) were admitted to the hospital because of
infectious diseases or respiratory distress syndrome, but
with a structurally normal heart, and underwent the same
detailed echocardiographic measurements. For this group
of patients, the measurement of d8 was left out, and d12 was
defined as the narrowest diameter of the isthmus of the
aorta.
Statistical Methods
Masked interobserver variability pertaining to echocar-
diographic measurements was not significant. Measure-
ments were recorded in millimeters and represent abso-
lute values. All data are presented as mean values and
standard deviations in parentheses. Windows Excel Ver-
sion 97 and the Statview 5.01 statistical program were
used for calculations and statistical analysis. Mean values
and standard deviations of demographic and echocardi-
ography data of both groups were compared with the
unpaired Student t test. Statistical significance was de-
fined as a p value of less than 0.05.
Results
Of the 47 neonates and 16 infants undergoing surgical
repair for coarctation, there was no surgical mortality.
Two neonates with severe aortic arch hypoplasia re-
quired early redo surgery for residual coarctation (3.2%),
and subsequently fared well. There was no morbidity in
the infant group. The data are hereafter regrouped and
presented for the 63 neonates and 23 infants.
The demographic and echocardiographic data of the 63
neonates are summarized in Table 1. Associated cardiac
defects in the group with coarctation (n ϭ 47) were as
follows: patent ductus arteriosus in 20 patients (42%), ven-
tricular septal defect in 20 patients (42%), bicuspid aortic
valve with or without aortic valve stenosis in 20 patients
(42%), and atrial septal defect or foramen ovale in 14
patients (30%). Two patients had chromosomal abnormali-
ties, 1 with Down syndrome, and 1 with Turner syndrome.
During the same period, echocardiographic measure-
ments from 23 infants were obtained; 16 of these under-
went surgery for repair of coarctation, and 7 belong to the
control group. The demographic and echocardiographic
data of these infants are summarized in Table 2. Associ-
ated cardiac defects in the group with coarctation were
bicuspid aortic valve with or without aortic valve stenosis
in 12 patients (75%), ventricular septal defect in 10
patients (63%), and patent ductus arteriosus in 3 patients
(19%). None of the controls had associated cardiac de-
fects or a PDA.
Great Vessel and Aortic Arch Dimensions
The diameters of the ascending and descending aorta
were not significantly different in patients with coarcta-
tion, neither for neonates nor for infants, as compared
with controls. The dimensions of the transverse arch
were significantly smaller in the coarctation group, espe-
cially in neonates. The distances between the origins of
the great vessels were larger in patients with coarctation
Fig 1. Scheme of a normal aortic arch (left) and of coarctation of the aorta (right). The following measurements were obtained: d1 ϭ proximal
ascending aorta diameter (measured at the level of the right pulmonary artery); d2 ϭ distal ascending aorta diameter (at the origin of the bra-
chiocephalic trunk); d3 ϭ proximal transverse arch diameter (at the origin of the left carotid artery); d4 ϭ distal transverse arch diameter (at
the origin of the left subclavian artery); d5 ϭ descending aorta diameter (distal to the isthmic region); d6 ϭ distance between the origin of the
brachiocephalic trunk and the origin of the left carotid artery; d7 ϭ distance between the origin of the left carotid artery and the origin of the
left subclavian artery; d8 ϭ distance between the origin of the left subclavian artery and the coarctation of the aorta; d9 ϭ diameter of the
origin of the brachiocephalic trunk; d10 ϭ diameter of the origin of the left carotid artery; d11 ϭ diameter of the origin of the left subclavian
artery; and d12 ϭ narrowest diameter of the coarctation.
1653Ann Thorac Surg DODGE-KHATAMI ET AL
2005;80:1652–8 CAROTID-SUBCLAVIAN ARTERY INDEX
CARDIOVASCULAR
3. than in controls, both in neonates and infants: the mean
distance from the brachiocephalic trunk to the carotid
artery (d6) in neonate patients with coarctation was 2.8
mm, compared with 1.5 mm in controls (p ϭ 0.0013). In
infants, the distance in patients with coarctation was 3.9
mm, compared with 2 mm in controls. The mean distance
from the left carotid artery (LCA) to the left subclavian
artery (LSA [d7]) in the neonate group with coarctation
was 7.32 mm, compared with 2.37 mm in neonate controls
(p Ͻ 0.0001). In infants, the mean distance from the LCA
to the LSA (d7) was 7.27 mm in those with coarctation,
compared with 2.67 mm in controls (p Ͻ 0.0001) (Fig 2).
The diameters of the great vessels were larger in the
coarctation group for neonates and infants; however,
significant increases were found in d10 only. Upon sub-
group analysis of patients with associated intracardiac
shunts or a PDA, there was no significant difference in
great vessel or arch dimensions, as compared with pa-
tients without associated defects.
To have a comparative parameter, we calculated the
ratios d1 to d7, d3 to d7, and d4 to d7. These indices were
proportionally significantly smaller in coarctation pa-
tients, when compared with either control neonates or
control infants (Table 3).
We used these ratios, d1/d7, d3/d7, and d4/d7, to find
predictive accuracy of two-dimensional echocardiogra-
phy in the diagnosis of coarctation for neonates, as well
Table 2. Demographic Data and Variables in Infants:
Coarctation and Controls
Infants
Coarctation
Patients
n ϭ 16
Controls
n ϭ 7 p Value
Demographic data
Age (days) 75 (34) 55 (12) 0.1318
Weight (kg) 4.43 (1.38) 4.45 (0.64) 0.9642
Length (cm) 56 (6) 55 (3) 0.6168
Body surface (m2
) 0.23 (0.07) 0.24 (0.02) 0.6967
Further measurements
Shortening fraction
of LV (%)
34 (6) 38 (5)
Gradient maximum
at COA (mm Hg)
48 (26)
Flow velocity
maximum at COA
(cm/s)
319 (116) 121 (12)
Aortic dimension
d1 (mm) 7.8 (1.1) 8.2 (2.2) 0.5576
d2 (mm) 6.8 (1.1) 7.4 (1.8) 0.2868
d3 (mm) 5.5 (1.4) 6.6 (0.8) 0.0639
d4 (mm) 4.5 (0.9) 6.3 (0.9) 0.0003
d5 (mm) 7.3 (1.9) 6.5 (0.8) 0.3054
d6 (mm) 3.9 (1.8) 2.0 (0.6) 0.0133
d7 (mm) 7.3 (2.4) 2.7 (0.8) Ͻ 0.0001
d8 (mm) 5.6 (2.5)
d9 (mm) 4.7 (1.2) 4.4 (0.5) 0.4945
d10 (mm) 3.3 (0.8) 2.4 (0.2) 0.0090
d11 (mm) 2.5 (0.5) 2.4 (0.2) 0.6439
d12 (mm) 2.3 (0.8) 5.6 (0.9) Ͻ 0.0001
Mean values are given, followed by standard deviation in parentheses.
COA ϭ coarctation; LV ϭ left ventricle.
Table 3. Ratios of Aortic Arch Dimensions to Great Vessel
Distances: Coarctation and Controls
Coarctation
Patients
n ϭ 63
Controls
n ϭ 23 p Value
Neonates 47 16
Index d1/d7 1.13 (0.83) 3.56 (1.55) Ͻ 0.0001
Index d3/d7 0.98 (0.87) 3.38 (1.43) Ͻ 0.0001
Index d4/d7 0.76 (0.86) 2.95 (1.24) Ͻ 0.0001
Infants 16 7
Index d1/d7 1.17 (0.43) 3.17 (0.83) Ͻ 0.0001
Index d3/d7 1.04 (0.43) 2.94 (0.88) Ͻ 0.0001
Index d4/d7 0.81 (0.29) 2.66 (0.78) Ͻ 0.0001
Mean values are given, followed by standard deviation in parentheses.
Table 1. Demographic Data and Variables in Neonates:
Coarctation and Controls
Neonates
Coarctation
Patients
n ϭ 47
Controls
n ϭ 16 p Value
Demographic data
Age (days) 12 (10) 16 (12) 0.15
Weight (kg) 3.0 (0.6) 3.2 (0.9) 0.37
Length (cm) 50 (7) 50 (4) 0.90
Body surface (m2
) 0.20 (0.02) 0.20 (0.04) 0.52
Further measurements
Shortening fraction
of LV (%)
34 (9) 36 (7)
Gradient maximum
at COA (mm Hg)
31 (18)
Flow velocity
maximum at COA
(cm/s)
267 (80) 130 (28)
Aortic dimension
d1 (mm) 6.8 (1.5) 7.5 (1.3) 0.0965
d2 (mm) 5.6 (1.1) 7.1 (1.2) Ͻ 0.0001
d3 (mm) 4.3 (1.0) 6.2 (1.3) Ͻ 0.0001
d4 (mm) 3.4 (0.8) 5.9 (1.4) Ͻ 0.0001
d5 (mm) 6.2 (1.4) 5.9 (1.1) 0.3227
d6 (mm) 2.8 (1.5) 1.5 (0.4) 0.0013
d7 (mm) 7.3 (3.0) 2.4 (0.8) Ͻ 0.0001
d8 (mm) 3.6 (1.6)
d9 (mm) 4.1 (0.9) 3.8 (1.1) 0.2494
d10 (mm) 2.8 (0.6) 2.4 (0.5) 0.0174
d11 (mm) 2.2 (1.2) 2.2 (0.4) 0.9052
d12 (mm) 2.1 (0.9) 5.0 (1.1) Ͻ 0.0001
Mean values are given, followed by standard deviation in parentheses.
COA ϭ coarctation; LV ϭ left ventricle.
1654 DODGE-KHATAMI ET AL Ann Thorac Surg
CAROTID-SUBCLAVIAN ARTERY INDEX 2005;80:1652–8
CARDIOVASCULAR
4. as for infants. To facilitate the recognition of coarctation,
we defined the index d4/d7 as the carotid-subclavian
artery index. This ratio was significantly smaller in the
coarctation group in neonates and infants, compared
with their respective controls. If the cut-off point for the
carotid-subclavian artery index is fixed at 1.5, there is a
sensitivity of 97.7% and a specificity of 92.3% for a
neonate to have coarctation, with a positive predictive
value of 97.7%, and a negative predictive value of 92.3%.
With a similar cut-off for the carotid-subclavian artery
index in infants, our data show a sensitivity of 94.7% and
a specificity of 100%. The positive predictive value is
100%, and the negative predictive value 90.9% (Table 4).
Regarding neonates only, an index d4/d7 below 2 gives a
very specific and sensitive result, but when infants are
included, a d4/d7 index below 1.5 gives the most accurate
results taking both age groups into consideration.
Comment
Since the early 1980s, the method of diagnosis for coarc-
tation has changed from using clinical data, with or
without preoperative catheter confirmation, to relying
almost exclusively on echocardiography [4]. Echocardi-
ography can allow noninvasive assessment of the aortic
arch, identification of the narrowing at the aortic isthmus,
flow measurement, and determination of the instant
gradient over the coarctation [5–7]. However, a signifi-
cant number of patients with coarctation are not properly
diagnosed during the neonatal period [5, 6]. That may be
due to patent ductus arteriosus without flow acceleration
at the isthmus of the aorta, to poor image quality, or to a
location further downstream in the descending aorta.
Furthermore, clinical judgment may be impaired in situ-
ations with diminished contractility of the left ventricle
and poor cardiac output, or other reasons such as infec-
tion or breathing artifacts [8]. Another potential problem
is, that even with the use of Doppler flow assessment in
the descending aorta, the anatomic severity of coarcta-
tion cannot always be assessed [2, 9–11]. Other authors
have tried to find a reliable echocardiographic parameter
to predict aortic coarctation in the newborn using mor-
phologic measurements, including aortic arch diameters
at different sites, calculations and comparison of diame-
ter ratios, or measurements of distances between the
great vessels of the aortic arch [12, 13]. That has to date
not given satisfying results to clearly identify a coarcta-
tion in difficult situations, and too many diagnoses have
gone unrecognized.
The study by Morrow and coworkers [12] enforces our
results, reporting significant alterations in the dimen-
sions of arch diameters, although by invasive angiogra-
Table 4. Sensitivity, Specificity, Positive and Negative Predictive Values According to Cut-Off
Sensitivity % Specificity % Positive Predictive Value % Negative Predictive Value %
Neonates
Index d1/d7 Ͻ 1.0 59.09 100.0 100.0 41.93
Index d1/d7 Ͻ 1.5 88.63 100.0 100.0 72.22
Index d1/d7 Ͻ 2.0 97.72 92.30 97.72 92.30
Index d1/d7 Ͻ 2.5 100.0 69.23 91.66 100.0
Index d3/d7 Ͻ 1.0 84.09 100.0 100.0 65.00
Index d3/d7 Ͻ 1.5 50.00 92.30 95.65 35.29
Index d3/d7 Ͻ 2.0 97.72 84.61 95.55 91.66
Index d3/d7 Ͻ 2.5 97.72 61.53 89.58 88.88
Index d4/d7 Ͻ 1.0 97.72 100.0 100.0 92.85
Index d4/d7 Ͻ 1.5 97.72 92.30 97.72 92.30
Index d4/d7 Ͻ 2.0 97.72 97.72 97.72 90.00
Index d4/d7 Ͻ 2.5 100.0 53.84 88.00 100.0
Infants
Index d1/d7 Ͻ 1.0 52.63 100.0 100.0 52.63
Index d1/d7 Ͻ 1.5 89.47 100.0 100.0 83.33
Index d1/d7 Ͻ 2.0 84.21 90.00 94.11 75.00
Index d1/d7 Ͻ 2.5 94.73 90.00 94.73 90.00
Index d3/d7 Ͻ 1.0 63.15 100.0 100.0 58.82
Index d3/d7 Ͻ 1.5 84.21 100.0 100.0 76.92
Index d3/d7 Ͻ 2.0 89.47 90.00 94.44 81.81
Index d3/d7 Ͻ 2.5 100.0 80.00 90.47 100.0
Index d4/d7 Ͻ 1.0 89.47 100.0 100.0 83.33
Index d4/d7 Ͻ 1.5 94.72 100.0 100.0 90.90
Index d4/d7 Ͻ 2.0 100.0 80.00 90.47 100.0
Index d4/d7 Ͻ 2.5 100.0 50.00 79.16 100.0
1655Ann Thorac Surg DODGE-KHATAMI ET AL
2005;80:1652–8 CAROTID-SUBCLAVIAN ARTERY INDEX
CARDIOVASCULAR
5. phy. They found no differences between patients and
controls concerning the descending aorta and left sub-
clavian artery diameters, but demonstrated that the
length of the transverse arch between the LCA and LSA
was significantly increased in patients with coarctation
[12]. Our results support his findings and add a useful
and reproducible index, with the use of a noninvasive
diagnostic tool. Nihoyannopoulos and associates [14]
assessed the predictive accuracy of two-dimensional
echocardiography in defining aortic arch obstruction.
Using viewing of the aortic arch only, the overall sensi-
tivity of the method was only 88%. They found two-
dimensional echocardiography to be more specific than
sensitive for the prediction of aortic arch obstruction,
noting that with a low origin of the LSA, particular
attention should be paid to the visualization of the
isthmus [14].
Contrary to our findings, Aluquin and coworkers [13]
found the distal ascending root diameter and descending
aorta to be significantly larger in patients with coarcta-
tion. Our data show that the proximal and distal diame-
ters of the ascending aorta are smaller in patients with
coarctation, and that the diameter of the descending
aorta is larger in coarctation patients, either due to
increased resistance before the stenosis or to post-
stenotic dilatation from turbulent flow. Nevertheless, our
data concur with theirs regarding the transverse arch,
which was notably longer in the coarctation group, as
compared with controls.
Excluding older invasive angiographic studies, newer
noninvasive modalities to accurately assess and diagnose
coarctation in the younger population exist, and are both
reliable and reproducible [2, 15]. These include axial,
multiplanar computed tomography scan and magnetic
resonance imaging, which are more expensive, cumber-
some, and could require anesthesia and intubation in the
newborn and infant population.
Because of the significant decrease in diameter of the
distal transverse aortic arch just before the LSA (d4) in
patients with coarctation, and the significant prolonga-
tion of the distance from the origin of the LCA to the
origin of the LSA (d7), we found it useful to use these two
variables as part of the carotid-subclavian artery index.
Therefore, we propose the carotid-subclavian artery in-
dex, where the diameter of the transverse arch at the
origin of the LSA (d4), is put in ratio to the distance from
the origin of the LCA to the origin of the LSA (d7), as a
screening tool for coarctation. In neonates and young
infants with coarctation, the carotid-subclavian artery
index yields a sensitivity of 97.7% for neonates and 94.7%
for infants, using a cut-off point below 1.5. The longer the
distance (d7) and the smaller the diameter of the aortic
arch at the origin of the LSA (d4), the smaller the
carotid-subclavian artery index, and the higher the pre-
dictability of coarctation. These findings remain valid
regardless of the presence or absence of an associated
intracardiac shunt or PDA.
Study Limitations
The results of our study are to be taken into the perspec-
tive of a retrospective design and its limitations. To
achieve validity, the carotid-subclavian artery index
should be prospectively assessed in patients with only
mild hypoplasia of the aortic arch, with or without
coarctation. Also, the numbers are relatively small, re-
ducing the power of the finding. To establish the useful-
ness of the carotid-subclavian artery index as a screening
tool for coarctation, a prospective study with a greater
population of newborns and infants is needed, both with
and without coarctation.
In conclusion, the carotid-subclavian artery index is
a simple screening parameter, readily obtained, and
standardized from two-dimensional echocardiography
visualization of the aortic arch. It shows high sensitiv-
ity and specificity for coarctation in our population of
newborns and infants with a cut-off point below 1.5,
independently of concomitant intracardiac or extracar-
diac shunts. In difficult subsets of patients with a large
PDA and severe concurrent illness with hemodynamic
instability, measuring the carotid-subclavian artery
Fig 2. Echocardographic images of two different aortic arches with a large distance between the left carotid artery and the left subclavian ar-
tery and significant narrowing of the transverse arch. Calculation of the carotid-subclavian index is highly specific for the presence of coarcta-
tion. (AAO ϭ ascending aorta; LCA ϭ left carotid artery; LSA ϭ left subclavian artery; TAA ϭ transverse aortic arch; Tr. brach. ϭ bra-
chiocephalic trunk.)
1656 DODGE-KHATAMI ET AL Ann Thorac Surg
CAROTID-SUBCLAVIAN ARTERY INDEX 2005;80:1652–8
CARDIOVASCULAR