Clinical Science (1997) 93, 3 17-324 (Printed in Great Britain)
317
Reproducibility of ultrasonographic measurements of different
carotid and femoral artery segments in healthy subjects and in
patients with increased intima-media thickness
T. J. SMILDE, H. WOLLERSHEIM, H. VAN LANGEN* and A. F. H. STALENHOEF
Departments of Medicine, Division of General Internal Medicine, University Hospital Nijmegen, PO Box
9 I0 I, 6500 HE3 Nijmegen, Nijmegen, The Netherlands, and *Clinical Vascular Laboratory, University
Hospital Nijmegen, PO Box 9/01, 6500 H B Nijmegen, The Netherlands
(Received 13 January/l9 June 1997; accepted 30 June 1997)
1. The reproducibility of measurements of the
arterial wall thickness in both the carotid and femoral artery was investigated by means of high-resolution B-mode ultrasonography. For this purpose,
subjects with normal and increased intima-media
thickness were selected. Images were stored on a n
optical disk and were analysed with a semi-automatic software program by two readers. Individuals
were scanned twice by two independent observers.
2. Measurements were performed of the far and
near wall of the common carotid artery and bulbous
in 30 healthy subjects and 19 patients known to
have a n increased intima-media thickness. Far-wall
measurements were made of the internal carotid
artery on both sides and common femoral artery on
the right side only.
3. In healthy subjects the mean within-observer
coefficient of variation was 1.8% and 3.0% for the
far wall in the common carotid artery on the right
side and left side, respectively. For the near wall the
mean coefficient of variation of the common carotid
artery was 2.8% on the right and 3.4% on the left
side. The mean coefficient of variation was less than
4% for both far and near wall in the bulbous and
far wall in the internal carotid artery. Even in
patients with increased intima-media thickness the
mean coefficient of variation of each segement was
less than 4.5%. In the control subjects the betweenobserver coefficient of variation of the common
carotid artery was 2.8% and 5.1% for the far wall on
the right and left side, respectively, and 3.4% and
4.2% for the near wall on the right and left side. In
healthy subjects a mean difference of 0.002mm
within observers was found in the right far-wall
common carotid artery, with limits of agreement of
-0.048 to 0.052 mm. The coefficient of repeatability
was 0.050mm. For patients with increased intimamedia thickness the mean difference in this segment
was -0.006 mm (-0.094 to 0.082) with a coefficient
of repeatability of 0.0i38mm. For the near wall in
the common carotid artery and far and near wall in
the bulbous and internal carotid artery the mean
differences were larger, but were all below 0.1 mm.
The differences and limits of agreements increased
between observers. In patients the between-observer
mean difference of the far wall of the common carotid artery was -0.055 mm (-0.255 to 0.145). For
the common femoral artery of normal control subjects the within- and between-observer mean differences were 0.005 mm (-0.119 to 0.129) and
0.015 mm (-0.081 to O.lll), respectively.
4. I n conclusion, the reproducibility of intimamedia thickness measurements in the common
carotid artery is reliable, even in patients with
increased artery wall thickness. Also in other segments prone to atherosclerosis, such as the bulbous,
internal carotid artery and common femoral artery,
a good reproducibility was found. To obtain good
reproducibility it is highly recommended to use the
same ultrasonographer to scan patients in follow-up
studies.
INTRODUCTION
Since the identification of lumen-intima and
media-adventitia echoes [l], high-resolution B-mode
ultrasonography has been used for noninvasive
quantitative measurements of intima-media thickness (IMT) which is associated with the presence of
atherosclerotic disease elsewhere [l-111. The assessment of IMT is an important tool in intervention
trials [12-171. Ultrasonogrpahy has several advantages over contrast anteriography: it can repetitively
be applied to asymptomatic subjects, it is relatively
cheap and safe and it measures both wall thickness
and lumen diameter.
The quality of ultrasonographic assessment of
atherosclerotic disease is highly dependent on the
Key words: atherosclerosis,carotid artery, femoral artery, intima-mediathickness, reproducibility,ultmonography.
Abbreviations: BUL,bulbous; CCA, common carotid artery; CFA, common femoral artery; CV,coefficient of variation; ICA, internal carotid artery; IMT intima-mediathickness;
RC. coefficient of repeatability.
Correspondence: Dr T. 1. Smilde.
318
T.J. Smilde et al.
instrumentation and the observer. Since ultrasonographic scanning cannot be automated, the observer
is an important source of measurement variability.
Most studies involving reproducibility and variability
report on far-wall measurements of the common
carotid artery only (CCA) and are mostly performed
in healthy subjects with normal IMT [2, 18, 191.
Although the femoral artery is a preferred site for
atherosclerosis, only few studies report on femoral
artery IMT [20-221, while reproducibility figures are
lacking.
The purpose of the present study was to invesigate the within- and between-observer variability
and reproducibility of measurements of the mean
IMT at different sites: the far and near wall of the
CCA, bulbous (BUL) and far wall of the internal
carotid artery (ICA) on both sides and far wall
measurements of the right common femoral artery
(CFA) in subjects with normal and increased IMT.
METHODS
This cross-sectional study was performed in 30
healthy subjects (50% male), without cardiovascular
risk factors except for smoking (n = 5 ) with an IMT
< 1.1mm, and in 18 male smokers known to have
intima-media thickening of 21.1 mm in at least one
segment. The latter were selected from a study of
various risk factors and vessel wall changes. All subjects were investigated twice with an interval
between 5 and 90 days. The first time observer 1
conducted the scanning, the second visit scannings
were made by two observers, blind for each other. In
four control subjects scannings were performed by
observer 1 on five consecutive days. The study was
approved by the Ethics Committee of our hospital.
Informed consent was obtained from each individual.
Ultrasound scanning protocol
The ultrasound examinations were performed
using a Biosound Phase 2 real-time scanner
equipped with a 10 MHz transducer. Two black and
white monitors displayed the B-mode ultrasound
images with spectrum analysis of the Doppler signals. Images were grabbed by a computer, stored on
a hard disk and analysed with a semi-automatic software program (Eurequa; TSA company, Meudon,
France) [18].
The scannings were performed with the subject in
a comfortable supine position, the head rotated
approximately 45 degrees away from the side being
scanned. The scanning sites involved were: the far
and near wall of the distal 1.0 cm of the straight part
of the CCA, the far and near wall of the carotid
bifurcation, beginning at the tip of the flow divider
and extending 0.8-1.0 cm proximal, and the far wall
of the proximal 1.0cm of the ICA. Measurements
were performed on the right and left side. Three
angles of interrogation were used: anterolateral,
lateral and posterolateral. In each individual the
most optimal scanning position (i.e. the head position and scanning angle which images the clearest
and thickest projection of interfaces) was noted on a
worksheet. The sonographer had the responsibility
of differentiating the ICA from external carotid
artery. This was accomplished by using several
criteria. Often the ICA is the furthest from the skin
surface when the V-shape of the flow divider is seen.
Other criteria are the larger luminal diameter of the
ICA and the dilatation in the ICA on the lateral or
posterolateral view. Apart from these characteristics
Doppler analysis was used to avoid misinterpretation.
Finally, scanning of the far wall of the right CFA
was performed 1 cm proximal to the descent of the
deep femoral artery. A fixed angle of insonation,
anteroposterior, was used. The scan converter
enabled freezing of the images during scanning. Callipers were placed on the anatomic references and
on the edges of the far and near wall. Subsequently,
the digitized frozen images with the clearest and
thickest projection were stored on disk. The worksheet with data on head position and scanning angle
was used for the second scanning and passed on to
the second observer.
Ultrasound analysis and report
The images stored on disk were read by two independent readers. Each segment was analysed separately. The reader selected the best measurable
portion of the image. Three measurements were
made in a preselected segment with a length of
0.5 cm. The measurements were performed automatically from significant changes in density on a
section perpendicular to the vessel wall from the
lumen towards subadventitial structures. The procedure was repeated over 0.5 cm adjacent to the
first portion and the mean thickness over 1 cm was
calculated. When it was not possible to measure the
IMT over the whole length of the selected segments,
for example in the bifurcation or when a plaque was
present, a smaller sample size was taken with a
minimum of 0.2cm. In neither case did the readers
have access to the IMT data of previous examinations. Measurements of the 12 individual segments
were noted. Missing data were scored.
Statistical analysis
Coefficients of variation (CV) were calculated as
the proportion of the SD of the mean. For CVs
describing the between-observer variability, SD was
computed over the first measurement of observer 1
and the measurement of observer 2. The CV was
estimated for each individual segment and of combinations of far and near wall per segment on the
right and left side and of the total mean combined
Reproducibility of arterial wall thickness
score of the 10 measurements in the carotid artery
(far and near wall of the CCA and BUL and far wall
of the ICA on left and right side). In addition, mean
differences between the first and second scanning of
observer 1 and the first scanning of observer 1 and
observer 2 were determined. By using the method of
limits of agreement as described by Bland and A t -
319
man [23], data are plotted in Figures 1 and 2, showing differences against the mean for each subject. In
addition, the coefficient of repeatability (RC) was
calculated for all segments as the SD of the estimated difference between two measurements,
assuming the mean difference to be zero. The 95%
confidence interval of the expected difference is cal-
0.15
g
.-
0,1
b
a
8
2
0.05
g
z o
L
-0.05
B
d
-0,15
-O'l
-0.2
4
I
032
0
44
0.6
0,s
1
1,2
1,6
1,4
Mean IMTof the far wall of the CCA
in mm
Fig. 1. Difference between the first and second scanning of observer I of the IMT of the far wall of the CCA plotted
against their mean. In this Figure the data of all subjects are included. The mean difference of 0.002 mm with the limits of agreement
(-0.048 to 0.052 mm) for healthy subjects is also indicated.
0.2
,
b
Mean IMT of the far wall of the CCA
in mm
Fig. 2. Difference between observer I and observer two of the IMT of the far wall of the CCA plotted against their mean.
In this Figure the data of all subjects are included. The mean difference of -0.004 mm with limits of agreements (-0.082-0.074) for
healthy subjects is also indicated.
T. J. Smilde et al.
320
results are listed of the measurements of the IMT of
the far wall of the right CCA, the BUL and CFA in
the 30 healthy control subjects. In Table 2 the CV
within and between observers in healthy control subjects and patients with increased IMT is given.
Although the CV remained low, the number of
missing data increased moving distally from CCA to
ICA (Table 3). Both observers visualized the near
and far wall of the CCA in all sessions. Of the far
and near walls in the bulb obtained in the first session, only 3% could not be found the second time by
the same observer. This increased to 18% when
compared with observer 2, meaning that 6% of the
far and 12% of the near walls obtained by observer
1 could not be visualized by observer 2 and vice
versa. Table 4 gives the absolute differences in mm
(SD) within and between observers, as measured in
the different segments of the carotid artery and the
CFA. In the control subjects the within-observer RC
ranged from 0.050mm for the far wall of the right
CCA to 0.070 mm for the far wall in the right CCA
(Table 5). The reproducibility of the combined
culated as 1.96 RC (definition adopted by the
British Standards Institution) [23], meaning that
repeated measurements are expected to differ by
more than the confidence interval with a probability
of only 5%. In the four subjects measured on five
consecutive days we calculated the SD of the serial
measurements for that subject against their mean
and then calculated the CV. Between-reader reproducibility was also expressed as CV and mean difference.
RESULTS
The mean age (SD) of the 30 control subjects was
45 (12) years. Fifteen subjects were male; two of
them smoked while three were past smokers. Mean
total serum cholesterol was 5.6 (1.3) mmol/l. All
participants with known thickened intima-media
were male smokers. Their mean age was 55 (8)
years and their mean total serum cholesterol concentration was 6.2 (1.5) mmol/l. In Table 1 the
Table 1. Detailed overview of measurement of the mean IMT of the far wall of the right CCA, BUL and CFA in 30 healthy subjects. -, Only measured by
observer I. Abbreviations: Obs I, observer I; Obs 2, observer 2.
CCA Far wall
Obs I
Subject
no.
I
2
3
4
5
6
7
8
9
10
II
I2
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
BUL Far wall
Obs 2
Obs 2
Obs I
Age
(Years)
First
time
Second
time
First
time
First
time
Second
time
32
32
58
33
28
32
38
69
44
66
38
31
26
58
38
33
36
64
42
45
68
38
49
49
50
45
50
41
63
44
0.43
0.6 I
0.44
0.55
0.83
0.61
0.54
0.66
0.87
-
0.5 I
0.8 I
0.72
0.96
0.70
0.53
0.98
0.95
0.80
0.6 I
0.53
0.65
0.90
I.oo
0.64
0.98
0.84
0.67
0.6 I
I.02
0.62
0.77
0.77
I.05
0.83
0.58
0.83
0.70
0.63
0.66
0.73
0.59
0.87
0.85
0.8I
0.73
1.01
0.69
I .03
0.80
0.64
0.6 I
I.oo
0.6 I
0.77
0.78
I.05
0.84
0.57
0.84
0.70
0.58
0.67
0.7 I
0.60
0.86
0.85
0.80
0.75
0.59
0.60
0.69
0.84
0.99
0.71
I.05
0.84
0.63
0.6 I
0.99
0.63
0.74
0.76
I.05
0.84
0.59
0.8 I
0.70
0.6 I
0.64
0.69
0.62
0.96
0.8 I
0.84
0.75
CFA Far wall
First
time
0.73
0.95
0.68
0.97
0.97
Obs I
First
time
Second
time
0.41
0.43
I.04
0.54
0.44
0.36
0.95
I.06
0.45
0.42
I.04
0.53
0.43
0.38
0.90
I.09
0.80
0.89
0.71
0.86
0.85
0.84
0.98
-
0.77
0.89
0.93
0.76
0.93
0.89
0.93
0.85
0.86
I.02
0.90
0.76
0.85
0.87
0.85
0.49
0.85
0.76
I .oo
0.99
0.78
0.86
0.89
0.8I
0.44
0.83
0.74
I.03
0.90
0.80
0.81
0.92
I .08
0.99
0.80
0.78
0.64
0.75
0.78
0.77
0.69
0.76
0.77
0.79
0.58
0.76
I.08
0.96
0.82
I.04
0.97
0.82
0.89
0.97
0.82
0.86
0.83
0.84
0.71
0.77
0.9 I
0.69
0.84
0.93
0.73
0.76
0.99
0.77
0.64
I.06
0.88
0.7 I
0.83
0.7 I
0.80
0.74
0.83
Obs 2
0.87
0.75
0.87
0.83
0.94
0.89
0.75
0.93
0.88
First
time
0.55
0.45
0.40
0.88
1.12
0.81
0.88
0.72
0.85
0.82
0.80
0.9 I
0.88
0.92
0.88
0.85
0.78
0.44
Reproducibilityof arterial wall thickness
32 I
Table 2. CVs (SD) of different segments in the carotid and femoral artery. CVs are given between and within observers for
subjects with normal IMT and for subjects with increased IMT.
Subjects with normal IMT
CCA far wall right
CCA far wall left
CCA near wall right
CCA near wall left
BUL far wall right
BUL far wall left
BUL near wall right
BUL near wall left
ICA far wall right
ICA far wall left
CFA right side
CCA far and near wall
on right and left side
Combination of 10 carotid
measurements (left and right)
Within observer
Between observer
Within observer
Betweenobserver
1.8 (I .8)
3.0 (4.0)
2.8 (3.7)
3.4 (3.8)
2.1 (1.8)
3.2 (4.4)
1.5 (1.3)
3.1 (4.7)
3.8 (5.1)
3.9 (4.4)
2.3 (2.1)
1.5 (1.6)
2.8 (2.4)
5.1 (6.2)
3.4 (3.7)
4.2 (4.4)
5.1 (5.7)
3.4 (2.6)
5.0 (3.3)
4.8 (6.0)
5.1 (4.2)
6.4 (4.9)
2.5 (2.5)
2.0 (I.9)
2.3 (2.4)
2.7 (4.6)
2.8 (6.0)
2.8 (3.6)
4.1 (5.2)
3.1 (1.7)
2.7 (3.7)
2.0 (I .3)
1.7 (1.8)
3.4 (2.4)
2.9 (I.9)
2.2 (3.7)
3.5 (5.0)
4.7 (5.9)
8.2 (9.9)
5.9 (8.1)
7.5 (9.2)
2.5 (0.5)
6.2 (4.3)
2.4 (1.1)
3.2 (2.6)
6.4 (I 0.5)
4.5 (10.5)
4.4 (4. I)
2.4 (2.8)
3.2 (3.5)
I.9 (2.0)
4.8 (5.0)
Table 3. Percentage of missing data in scannings made by observer I
(n = 96) and observer 2 (n = 45)
Missingdata (%)
Observer I,
first time
Observer I,
second time
0
0
31
60
42
0
0
31
58
43
62
38
54
Observer 2,
first time
~
CCA far and near wall right
CCA far and near wall left
BUL far wall right
BUL far wall left
BUL near wall right
BUL near wall left
ICA far wall right
ICA far wall left
64
38
53
Subjects with increased IMT
0
I
43
56
42
65
38
56
measurements of the far and near wall on both sides
of the CCA were better than the individual
measurements, except for the between-observer variability in patients with increased wall thickness
(Table 2). The CV of the total carotid artery, combining all measurements, was also low (betweenobserver CV, 3.2 and 4.8% for healthy subjects and
patients respectively, see Table 2). The CV of the
measurement of the far wall of the CFA was <5%
although the between-observer reproducibility was
clearly less (CV 4.4) than the within-observer reproducibility (CV 2.2) in subjects with increased IMT.
Fig. 1 shows the limits of agreement of the IMT
of the CCA measured by observer 1 in all subjects.
In this Figure the differences between the first and
second session are plotted against their mean. Fig. 2
shows the difference between measurements made
Table 4. Absolute differences in mm (SD)within and between-observers, as measured in the different segments of the
carotid artery and the far wall of the CFA in subjects with and without intima-media thickening
Subjects with normal IMT
CCA far wall right
CCA far wall left
CCA near wall right
CCA near wall left
BUL far wall right
BUL far wall left
BUL near wall right
BUL near wall left
ICA far wall right
ICA far wall left
CFA right side
CCA far and near wall
on right and left side
Combination of 10 carotid
measurements(left and right)
Subjects with increased IMT
Within observer
Between observer
Within observer
Between observer
0.017 (0.017)
0.026 (0.035)
0.031 (0.042)
0.026 (0.035)
0.025 (0.020)
0.041 (0.044)
0.023 (0.017)
0.040 (0.058)
0.032 (0.037)
0.030 (0.030)
0.03 I (0.03 I)
0.016 (0.018)
0.027 (0.022)
0.045 (0.055)
0.041 (0.042)
0.045 (0.055)
0.064 (0.072)
0.042 (0.034)
0.070 (0.042)
0.049 (0.070)
0.055 (0.046)
0.054 (0.043)
0.036 (0.036)
0.026 (0.025)
0.033 (0.029)
0.058 (0.I 15)
0.059 (0.152)
0.045 (0.045)
0.079 (0.099)
0.058 (0.030)
0.055 (0.081)
0.036 (0.025)
0.030 (0.050)
0.045 (0.030)
0.048 (0.043)
0.021 (0.019)
0.063 (0.101)
0.08 I (0.105)
0.140 (0.190)
0.100(0.163)
0.100 (0.140)
0.050 (0.008)
0. I26 (0.096)
0.040 (0.016)
0.062 (0.083)
0. I33 (0.298)
0.089 (0.218)
0.077 (0.076)
0.022 (0.026)
0.03 I (0.029)
0.036 (0.060)
0.072 (0.071)
T. J.Smilde et al.
322
Table 5. Mean differences in mm with (limits of agreements) within and between observers measured in the different
segments of the carotid and CFA in 30 healthy subjects and 18 subjects with increased IMT
Subjects with increased IMT
Subjects with normal IMT
Within observer
CCA far wall right
CCA far wall left
CCA near wall right
CCA near wall left
BUL far wall right
BUL far wall left
BUL near wall right
BUL near wall left
ICA far wall right
ICA far wall left
CFA right side
CCA far and near wall
on right and left side
Combination of 10 carotid
measurements (left and right)
0.002
(-0.048 to 0.052)
-0.0012
(0.102 to 0.078)
-0.013
(-0.I I3 to 0.087)
-0,009
(-0.133 toO.115)
0.002
(-0.072 to 0.068)
-0.01 I
(-0.133 to 0.1 I I)
0.009
(-0.043 to 0.061)
-0.036
(-0. I58 to 0.086)
-0.014
(-0.1 14 to 0.086)
0.0 I2
(-0.082 to 0. I 18)
0.005
(-0.119to0.129)
-0.006
(-0.052 to 0.040)
0.002
(-0.066 to 0.070)
by observer 1 and 2 plotted against their mean.
Table 5 shows the difference between measurements
with their limits of agreement. For those control
subjects measured on five consecutive days the CV
did not change over time; it ranged from 2.1% in
the CCA to 4.2% in the BUL.
All observations were read by two different readers. The inter-reader variability was less than 2% in
all segments, with a mean difference of lower than
0.015 mm. Variability did not differ between images
of normal control subjects and patients with
increased IMT.
DISCUSSION
With the ultrasound protocol used in this study it
was possible to visualize and measure IMT in different segments of the carotid artery, including in
patients with known increased IMT. The anatomical
location of a biological structure is always defined by
a leading edge of an echo, and the thickness of a
structure as the distance between the leading edges
of two different echoes. It has therefore been
argued that, in spite of the similarities of near- and
far-wall images, IMT can only be measured accurately in the far-wall position, because only the farwall IMT is defined by leading edges [24, 251.
Because of the echogenicity of the adventitia in the
Between observer
Within observer
Between observer
-0.004
-0.006
(-0.094
0.030
(-0.222
- 0.044
(-0.358
-0.006
(-0.128
-0.017
(-0.273
t00.I 16)
-0.055
(-0.255
0.022
(-0.242
-0.028
(-0.502
-0.05
(-0.422
to 0.239)
0.001
(-0.55 I to 0.553)
(-0.082 to 0.074)
- 0.003
(-0.151 t00.145)
- 0.007
(-0.117to0.103)
-0.1 15
(-0.259 to 0.029)
0.033
(-0.151 to 0.217)
-0.010
(-0.I20t00.100)
0.012
(-0. I56 to 0.180)
-0.032
(-0.202 to 0.138)
-0.018
(-0.163 t00.146)
0.036
(-0.086 to 0.158)
0.0 I5
(-0.081 toO.l I I)
-0.006
(-0.06 to 0.048)
-0.014
(-0.094 to 0.066)
to 0.082)
to 0.282)
to 0.270)
- 0.00 I
(-0.141 t00.139)
-0.038
(-0.222 to 0.146)
-0.021
(-0.101 to0.059)
-0.023
(-0.131 t00.106)
-0.005
(-0.121 to0.III)
-0.002
(-0.146 to 0.142)
0.009
(-0.131 t00.149)
-0.003
(0.09 I to 0.085)
to 0.145)
to 0.286)
to 0.446)
to 0.322)
0.024
'
(-0.062 to 0.1 10)
0.022
(-0.316 to 0.360)
-0.004
(-0.098 to 0.090)
-0.051
(-0.233 to0.131)
0.095
(-0.533 to 0.723)
-0.048
(-0.5I2t00.416)
-0.036
(-0.228 to 0.156)
-0.019
(-0.249 to0.21 I)
near wall, the reflections from the intima-media may
be blurred. However, by scanning in such a way that
the jugular vein is placed adjacent to the carotid
artery and with the development of new software it
was possible to measure near walls reliably, which
may provide additional information. In this study we
measured both far and near wall with the aid of a
semi-automatic analysing program in a standardized
way. Using the clearest and thickest projection of
the IMT leads to bias towards thicker values. This
scanning procedure was performed in order to see
whether measurements of increased IMT are reproducible. The mean differences between and within
sonographers were small, although from the limits
of agreement we concluded that, particularly in
patients with increased IMT, the between-observer
variation is larger in an individual patient. How far
apart measurements may be without causing difficulties is a matter of judgement and should be defined
in advance. The limits of agreement are only estimates of the values which apply to the whole population and one should realize that a different sample
would give different limits.
The reproducibility of the combined IMT of the
far and near wall of the CCA was better than the
individual measurements. This is probably due to
averaging more measurements. The CV was somewhat larger on the left than on the right side. This
might be due to the different anatomical situation or
Reproducibility of arterial wall thickness
by the right-handiness of the observers, which has
also been mentioned by others [19]. The CV and
mean differences did not change between consecutive days, indicating that the scanning method is reliable and repeatable.
The between-reader variation was small. In this
protocol the sonographer determined the best
image, which was stored on disk. Consequently, the
readers’ influence was less prominent since he was
confronted with a preselected image. The storage of
images on optical disks instead of taping them on
video has the advantage of full resolution preservation. Measurement error of IMT tended to increase
with increasing levels of IMT, as reported earlier
[19]. This study shows that also in segments other
than the CCA, IMT can be measured with great
reliability, even in patients with known increased
IMT. This is important, since intervention trials will
most often be performed in patients in which an
increased IMT is anticipated. Moving from the CCA
to the ICA, the amount of missing data increased
mostly because the site of the bifurcation was
located high up in the neck. Other studies encountered the same problem [5]. It is important to
realize that reliable, repeatable measurements in the
BUL and ICA are not possible in all patients. In calculating numbers of patients required for clinical
trials with IMT one should correct for these
expected missing data. Another important finding in
this study was that the same segments could be
visualized twice by the same observer in almost all
cases. Since the within-observer variability is smaller
than the between-observer variability and two observers are not always able to visualize the same segments, it is highly recommended that patients
should be scanned during follow-up by the same
observer.
In several studies the reproducibility of the IMT
has only been determined in the CCA in healthy
subjects [2, 18, 191. The CV in the study of Salonen
et al. [2] was approximately 5% and the mean difference in IMT between observers was 0.087mm. In
the multi-centre Cardiovascular Health Study the
between-sonographers difference was 0.20 [7]. In
both studies the differences between observers was
based on two and one measurements, respectively,
in contrast to six measurement sites in our study.
Furthermore, more observers were taken into
account, while in our study two experienced observers performed all measurements. The use of an
optical disk instead of video tapes makes the readers’ influence less and this may decrease the variability. The latter may explain the better results in
our study compared with those of Riley et al. [26].
They report on the reproducibility of combined
carotid artery wall thickness. The mean absolute difference within and between observers was 0.06 and
0.12mm, compared with 0.04 and 0.07mm in our
study in patients with increased IMT. Another
explanation for this difference may be the fact that
the former authors included the near wall of the
323
ICA. The reproducibility data of the Rotterdam
study are very similar to our data: a mean difference
of 0.013mm on the right and 0.05 on the left side
between sonographers at different visits in the CCA
~91.
The variability of measurements is determined by
the sonographer, reader, instrumentation and by differences between subjects. Especially in patients
with increased IMT of plaques, measurements
become more imprecise, because of tortuous arteries, eccentric plaque and irregularities. Since these
factors cannot be influenced, it is important to
reduce the effect of other factors determining variability. This means regular check-ups of the instrumentation, a repeated training programme of the
sonographers and readers with quality control
assessment and a follow-up schedule that allows
scanning by the same sonographer.
High-resolution B-mode ultrasonographic measurement of carotid arterial IMT is a suitable
pseudo end-point in clinical trials [13-17, 27, 281,
although the relation between IMT reduction and
clinical events as coronary heart disease and stroke
need to be established. The non-invasive imaging of
the arterial wall can be performed repeatedly in
symptomatic and asymptomatic patients, carried
negligible risk and quantifies early atherosclerosis
and atherosclerotic changes due to risk factor modification.
Atherosclerotic changes in the carotid artery are
not equally distributed. Increased IMT occurs more
often in the carotid BUL than in the CCA. As
shown by this study measurements at other sites
than the CCA can be done with good reproducibility
even in patients with thickened IMT.
In conclusion, measurements of the IMT in different segments of the carotid artery and of the femoral artery were highly reproducible. Although the
measurement error tended to increase with
increased mean IMT, the reproducibility remained
good. The reproducibility at the site of the BUL and
ICA were good whenever obtained, although it was
not always possible to visualize these segments. With
the possibility of direct storing images on optical
disk, the resolution is preserved. The betweenreader variability is very low when using a (semi)
automatic reading system. In follow-up studies it is
highly recommended that the same ultrasonographer be used to scan patients.
REFERENCES
I. Pignoli P, Tremoli E, Poli A, Oreste P, Paoletti R lntimal plus medial thickness of
the arterial wall: a direct measurement with ultrasound imaging. Circulation
1986; 7 4 1399-406.
2. Salonen R Haapanen A Solonen IT. Measurement of intima-media thickness of
common carotid arteries with high-resolution B-mode ultrasonography: interand intra-observer Variability. Ultrasound Med Biol 1991; 17: 225-30.
3. Penson J, Stavenow L. Wikstrand J, lsraelsson B, FormgrenJ, Berglund G.
Noninvasive quantification of atherosclerotic lesions. Reproducibilityof
ultrasonographic measurement of arterial wall thickness and plaque size.
Arteriosclerosis Thromb 1992; I2 26 1-6.
324
T. J. Smilde et al.
4. Okin PM, Roman MJ, Devereux RB, Kligfield P. Association of carotid
atherosclerosis with electrocardiographicmyocardial ischemia and left ventricular
hypertrophy. Hypertension 1996; 28: 3-7.
5. Salonen JT, Salonen R Ultrasonographically assessed carotid morphology and the
risk of coronary heart disease. Arteriosclerosis Thromb I99 I; I I : 1245-9.
6. Heiss G, Sharrett AR, Barnes R, Chambless LE, Szklo M, Alzola C. Carotid
atherosclerosis measured by B-mode ultrasound in populations; associations with
cardiovascular risk factors in the ARK study. Am J Epidemiol 1991; 134 250-6.
7. OLeary DH, Polak IF, Wolfson SKI, et al. Use of sonography to evaluate carotid
atherosclerosis in the elderly. The Cardiovascular Health Study. CHS
Colloborative Research Group. Stroke I99 I; 2 2 I 155-63.
8. Wendelhag I,Wiklund 0, Wikstrand J. Arterial wall thickness in familial
measurement of intima-media thickness in the common carotid artery.
Arteriosclerosis Thromb 1992; I 2 70-7.
9. Bond MG, Barnes RW, Riley WA, et al. High-resolution B-mode ultrasound
scanning methods in the atherosclerosis risk in communities study (ARIC). Am
Soc Neuroimag 1991: 68: 73.
10. Poli A, Tremoli E, Colombo 4 Sirtori M, Pignoli P. Paoletti R Uitrasonographic
measurement of the common carotid artery wall thickness in
hypercholesterolemicpatients. A new method for the quantitation and follow-up
of preclinical atherosclerosis in living human subjects. Atherosclerosis 1988; 70:
253-61.
I I. Bonithon Kopp C, Scarabin PY, Taquet A, Touboul PJ, Malmejac A, Guize L. Risk
factors for early carotid atherosclerosis in middle-aged French women.
Arteriosclerosis Thromb I99 I ; I I: 966-72.
12. Furberg CD, Borhani NO, Byington RP, Gibbons ME, Sowers JR Calcium
antagonist and atherosclerosis. Multicenter Isradipine/DiureticAtherosclerosis
Study. Am J Hypertens 1993; 6 245-95.
13. Hodis HN, Mack WJ, LaBree L, et al. Reduction in carotid arterial wall thickness
using lovastatin and dietary therapy. A randomized, controlled clinical trial. Ann
Intern Med 1996; I 2 4 548-56.
14. Kroon AA, Asten WNJC, Stalenhoef AFH. Effects of apheresis of low-density
lipoprotein on peripheral vascular disease in hypercholesterolemicpatients with
coronary artery disease. Ann Intern Med 1996; 125: 945-54.
15. Blankenhorn DH, Selzer RH, Crawford DW, et al. Beneficial effects of
colestipol-niacin therapy on the common carotid artery. Two- and four-year
reduction of intima-mediathickness measured by ultrasound. Circulation 1993;
88: 20-8.
16. Crouse JR, Byington RP, Bond MG. Pravastatin, lipids, and atherosclerosis in the
carotid arteries (PLAC-11). Am J Cardiol 1995; 75: 455-9.
17. de Groot E, JukemaJW, van Boven AJ, et al. Effect of pravastatin on progression
and regression of coronary atherosclerosis and vessel wall changes in carotid and
femoral arteries: a report from the Regression Growth Evaluation Statin Study.
Am J Cardiol 1995; 7 6 4OC-6C.
18. Touboul PJ. Prati P. Scarabin PY, Adrai V, Thibout E, Ducimetiere P. Use of
monitoringsoftware to improve the measurement of carotid wall thickness by
B-mode imaging. J Hypertens 1992; 10 (Suppl.): 537-41.
19. Bots ML, Mulder PG, Hofman A, van Es GA, Grobbee DE. Reproducibilityof
carotid vessel wall thickness measurements. The Rotterdam Study. J Clin
Epidemiol 1994; 47: 921-30.
20. Gariepy J, Simon 4 Massonneau M, Linhart A, Segond P, Levenson J, Groupe
PCVMETRk Echographic assessment of carotid and femoral structure in men
with essential hypertension. Am J Hypertens 1996; 9: 126-36.
21. Beach KW, Isaac CA. Phillips DJ, StrandnessJrE. An ultrasonic measurement of
superficial femoral artery wall thickness. Ultrasound Med Biol 1989; 15: 723-8.
22. JoensuuT, Salonen R Winblad I, Korpela H, Slaonen JT. Determinants of femoral
and carotid artery atherosclerosis. J Intern Med 1994; 236 79-84.
23. Bland JM. Altman DG. Statistical methods for assessing agreement between two
methods of clinical measurement. Lancet 1986; k 307-10.
24. Wendelhag I, Gustavsson T, Suurkula M. Berglund G, Wikdstrand J. Ultrasound
measurement of wall thickness in the carotid artery: fundamental principles and
description of a computerized analysing system. Clin Physiol 1991; I I: 565-77.
25. Wong M, EdelsteinJ,Wollman J, Bond MG. Ultrasonic-pathologicalcomparison
of the human arterial wall. Verification of intima-media thickness. Arteriosclerosis
Thromb 1993; I 3 482-6.
26. Riley WA, Barnes RW, Applegate WB, et al. Reproducibilityof noninvasive
ultrasonic measurement of carotid atherosclerosis. The asymptomatic carotid
artery plaque study. Stroke 1992; 23: 1062-8.
27. Blankenhorn DH, Hodis HN. George Lyman Duff Memorial Lecture. Arterial
imaging and atherosclerosis reversal. Arteriosclerosis Thromb 1994; 1 4 177-92.
28. Furberg CD, Byington RP, Borhani NA. Multicenter isradipine diuretic
atherosclerosis study (MIDAS). Design features. The Midas Research Group. Am
J Med 1989; 8 6 37-9.