Download Clinical predictors of the defibrillation threshold with the

1576
JACC Vol. 25, No. 7
June 1995:1576-83
Clinical Predictors of the Defibrillation Threshold With the Unipolar
Implantable Defibrillation System
MERRITT
H. R A I T T , M D , G E O R G E
JOHNSON,
B S E E , G. L E E D O L A C K , M D , F A C C ,
J E A N N E E. P O O L E , M D , F A C C , P E T E R J. K U D E N C H U K ,
M D , F A C C , G U S T H. B A R D Y , M D , F A C C
Seattle, Washington
Objectives. The purpose of this study was to determine the
relation between clinical variables and the defibrillation threshold
by using a standardized testing protocol and a uniform implantable defibrillator system.
Background. Past studies have not revealed useful correlations
between clinical variables and the energy required to terminate
ventricular fibrillation. Most of these studies did not use a
uniform implantable defibrillator system or a standardized protocol to measure the defibrillation threshold and, thus, did not
control for the influence of these technical influences. We postulated
that defibrillator and defibrillation threshold measurement-based
sources of variability overshadowed important clinical predictors.
Methods. The defibrillation threshold was measured by using a
standardized protocol in 101 consecutive patients. We used a
transvenons unipolar pectoral defibrillation system that employed
a single endocardial right ventricular defibrillation coil as the
anode and the shell of an 80-cm3 pulse generator as the cathode to
deliver a 65% tilt biphasic pulse.
Results. Several clinical variables were found to be significantly
associated with the defibrillation threshold: patient gender,
height, weight, body surface area, heart rate at rest, QRS and
corrected QT (QTc) intervals, left ventricular mass and several
measures of heart and chest size by chest roentgenogram. None of
these variables had a correlation coefficient >0.45 with the
defibrillation threshold. On multivariate analysis, left ventricular
mass and heart rate at rest were the only independent predictors
of the defibrillation threshold and explained only 25% of the
observed variability.
Conclusions. Despite the use of a uniform transvenous defibrillation system and a standardized protocol to measure the defibrillation threshold, no clinically relevant correlation was found
between clinical variables and the defibrillation threshold. The
defibrillation threshold is probably a function of a complex
interaction of anatomic, physiologic and cellular variables that
are not adequately represented by easily obtainable clinical
information. It is probably not possible to predict defibrillation
outcome from standard clinical variables.
Past efforts to relate the minimal energy required to terminate
ventricular fibrillation to clinical variables in patients undergoing defibrillator implantation have revealed poor correlations (1-5). If the defibrillation threshold could be predicted
on the basis of clinical data, such information could be useful
in planning patient care.
Many device and lead system factors are known to effect the
defibrillation threshold, including the number of electrodes
utilized, electrode position, electrode polarity, pulsing technique and shock waveform (4,6-17). It is common practice to
test a variety of combinations of these variables at the time of
defibrillator implantation in an attempt to optimize the defibrillation threshold of an individual patient (1-5,18,19). It was
our hypothesis that the wide patient to patient variation in
defibrillation technique in previous studies may have overshadowed the role of clinical variables in determining the
defibrillation threshold. Furthermore, not all past investigators
(2-5) have used a standardized protocol to determine the
defibrillation threshold and many (2,4,5) simply examined
whether a clinical variable predicted whether the defibrillation
threshold was above or below a predetermined arbitrary
energy level. This lack of precision in determining the defibrillation threshold, the variation in defibrillator systems and the
use of an arbitrary energy level instead of the defibrillation
threshold as an end point may have clouded the effect of
clinical variables on the defibrillation threshold. Thus, it was
the purpose of this study to determine the relation of clinical
variables to the defibrillation threshold in a large group uf
patients by using a uniform defibrillation protocol, lead system,
lead location, electrode polarity and shock waveform.
From the Division of Cardiology,Department of Medicine, Universityof
Washington, Seattle, Washington.This work was supported in part by a grant
from the National Heart, Lung, and Blood Institute, National Institutes of
Health, Bethesda, Maryland;MedtronicCorporation,Minneapolis,Minnesota;
and the TachycardiaResearch Foundation, Seattle, Washington.Dr. Bardyis a
consultant for Medtronics, Inc., Minneapolis.
Manuscript receivedOctober3, 1994;revised manuscriptreceivedDecember 30, 1994,acceptedFebruary2, 1995.
Address for correspondence:Dr. Merritt H. Raitt, PortlandVeteransAffairs
Medical Center, 3710SW US Veterans Road, P.O. Box 1034(IllB), Portland,
Oregon 97207.
©1995 by the AmericanCollegeof Cardiology
(J Am CoU Cardiol 1995;25:1576-83)
Methods
Patients. After providing informed verbal and written consent, 101 consecutive patients with a history of syncopal
ventricular tachycardia, ventricular fibrillation, or both, under0735-1097/95/$9.50
0735-1097(95)00093-J
JACC Vol. 25, No. 7
June 1995:1576-83
went defibrillation threshold testing with a unipolar single lead
transvenous defibrillation system during implantation of a
standard transvenous defibrillator.
Unipolar defibrillation system. The unipolar defibrillation
system utilized in this study has been described previously (20).
The anodal electrode is a 5-cm long coil on a 10.5F transvenous lead (Medtronic model 6966) positioned in the right
ventricular apex. The cathode is an active titanium shell with a
volume of 80 cm3 and a surface area of 108 cm2 (Medtronic
model 7219C) placed in a left infraclavicular pocket (19). This
unipolar defibrillation system utilizes a 120-~F capacitor to
deliver a 65% tilt asymmetric biphasic defibrillation pulse (15).
All defibrillation pulse characteristics were measured from
oscilloscopic recordings of waveform, voltage and current, as
previously described (13).
Defibrillation threshold testing. The defibrillation threshold of the unipolar system was determined before testing and
implantation of a standard system that utilized two transvenous
electrodes and a left infraclavicular subcutaneous patch. Standard antiarrhythmic medications were stopped at least 5
half-lives before testing, and amiodarone was stopped 1 to 2
weeks before testing. The energy was delivered 10 s after the
onset of ventricular fibrillation, including the time period
during which alternating current was applied to induce the
rhythm. If the transvenous pulse delivered was unsuccessful, a
100- to 200-J transthoracic rescue pulse was delivered immediately by means of a precharged external defibrillator (PhysioControl LifePak 6s). A minimal rest period of 3 min was
observed between inductions of ventricular fibrillation. Before
reinduction of ventricular fibrillation, care was taken to ensure
that the ST segments, T waves and arterial pressure had
returned to baseline level. The initial energy level tested was
10 J. The pulse energy was increased after a failed shock and
decreased after a successful shock. Testing then continued in
5-J steps between energy levels of 10 and 30 J, in 2.5-J steps
between 5 and 10 J and in 1.25-J steps at <5 J. If a shock was
successful the energy was decreased two steps for each retest
until a failure occurred, after which the next highest previously
skipped energy level was tested. Similarly, if the initial 10-J
energy level failed, it was increased two steps after each
ensuing failure and then reduced one step to the last skipped
step after a success. This skip method was used to minimize the
number of episodes of ventricular fibrillation needed to determine the defibrillation threshold. The defibrillation threshold
was defined as the lowest pulse amplitude that successfully
terminated ventricular fibrillation. Determining this defibrillation threshold required 2 to 5 (mean + SD 3.4 _+ 0.9)
inductions of ventricular fibrillation. All patients completed
the protocol without significant complications.
Clinical data. For each patient the clinical data available
included age, gender, height, weight, body surface area, clinical
arrhythmia history, underlying cardiac disease, chronic therapy
with amiodarone within the last month, past cardiac surgical
history, ejection fraction by radionuclide ventriculogram and
electrocardiographic (ECG) data including QRS interval, corrected QT (QTc) interval and heart rate at rest. A postero-
RAITT ET AL.
DEFIBRILLATION THRESHOLD PREDICTION
1577
Table 1. BaselinePatient Characteristics (n = 101)
Age (yr)
Male
Coronary artery disease
History of ventricular tachycardia
Past heart surgery
Past amiodarone therapy
Left ventricular ejection fraction
57 + 13
73%
68%
54%
47%
21%
0.38 + 0.16
Data are expressed as mean value _+ SD or percent of patients.
anterior chest roentgenogram performed after defibrillator
implantation was examined and several measures of heart size
and the relative positions of the pectoral and right ventricular
electrodes were recorded. These measurements included the
horizontal heart and chest diameters at the level of the cardiac
apex, the distance from the right ventricular electrode to the
center of the pectoral patch (which occupies the same pectoral
pocket occupied by the active can during testing for this study),
the distance along the line between the two defibrillation
electrodes and from the middle of the right ventricular coil to
the superior heart border, and the horizontal distance from the
chest midline to the center of the pectoral pocket (position of
the can electrode during testing). In 86 patients, left ventricular end-diastolic diameter, septal wall thickness and left ventricular free wall thickness were measured from echocardiograms performed as part of the patients' evaluation before
defibrillator implantation. Left ventricular mass was calculated
(21) from these measured variables.
Statistical analysis. The two-tailed t test and linear regression were used to analyze the statistical significance of correlations between the defibrillation threshold and clinical variables. Stepwise multiple linear regression (SPSS version 6.0)
with a probability of F to enter of 0.05 and 0.10 to remove was
used to determine independent predictors of the defibrillation
threshold.
Results
The baseline characteristics of the 101 patients in the study
group are shown in Table 1. The mean defibrillation threshold
was 9.3 _+ 6.5, with 98% of patients having a defibrillation
threshold <25 J. The correlation coefficients for the clinical,
chest X-ray and echocardiographic variables with the defibrillation threshold are shown in Table 2. Figures 1 to 4 show
graphs of the defibrillation threshold as a function of each of
the individual clinical variables. A statistically significant correlation was present between the defibrillation threshold and
several variables including patient gender, weight, body surface
area, heart rate at rest, QRS interval, QTc interval, chest
diameter, heart diameter, ejection fraction, left ventricular
end-diastolic diameter and left ventricular mass. Despite the
significance of the relation between these clinical variables and
the defibrillation threshold, the correlation coefficients were
poor, with none >0.45. Furthermore, when the data from the
two outlier patients with the highest defibrillation thresholds
1578
RAITI" ET AL.
DEFIBRILLATION T H R E S H O L D PREDICTION
JACC Vol. 25, No. 7
June 1995:1576-83
Table 2. Correlation Between Clinical Variables and the
Defibrillation Threshold, Univariate Analysis
were removed from the analysis, the QTc interval and ejection
fraction no longer had a significant association with the
defibrillation threshold. Left ventricular mass had the best
correlation with the defibrillation threshold (R = 0.45). There
were no significant associations between the defibrillation
threshold and patient age, history of heart surgery, history of
myocardial infarction, history of ventricular tachycardia, recent
amiodarone therapy, cardiothoracic ratio, the distance between the two electrodes on the posteroanterior chest roentgenogram, posterior and septal wall thickness on echocardiography and the resistance between the coil and can electrode.
Stepwise multiple linear regression was performed using
the variables with a significant association with the defibrillation threshold on univariate testing; 17 patients were excluded
because of missing data. Only two clinical variables were found
to be independent predictors of the defibrillation threshold:
left ventricular mass and heart rate at rest. The model containing both these variables nevertheless was a poor predictor
of the defibrillation threshold. The correlation coefficient was
only 0.50 and together these variables explained only 25% of
the observed variability in the defibrillation threshold.
Figure 1. Plots of the defibrillation threshold (DFF) as a function of
(A) gender, (B) previous history of coronary artery disease (CAD), (C)
previous history of myocardial infarction (MI), (D) previous history of
ventricular tachycardia (VT), (E) recent amiodarone therapy and (F)
previous open heart surgery. The p values are shown for a two-tailed
t test comparing the groups.
• A) p = 0 . 0 0 4
40
B) p = 0 . 0 4
Clinical Characteristic
Regression
Coefficient
p
Value
Age
Female
Height
Body weight
Body surface area
History of myocardial infarction
History of coronary artery disease
History of ventricular tachycardia
Past heart surgery
Recent amiodarone therapy
Heart rate at rest
QRS interval
QTc interval
PA chest width on chest roentgenogram
PA heart width on chest roentgenogram
Cardiothoracic ratio
Distance from RV coil to can electrode
Distance from RV coil to heart border
Distance from spine to can electrode
Ejection fraction
Posterior wall thickness
Septal wall thickness
Left ventricular end-diastolic diameter
Left ventricular mass
Shock resistance
+0.04
-0.28
+0.29
+0.33
+0.35
+0.14
+0.20
+0.15
+0.01
+0.12
+0.24
+0.26
+0.20
+0.29
+0.36
+0.09
+0.10
+0.29
+0.15
-0.25
+0.13
+0.03
+0.40
+0.45
+0.03
0.66
0.004
0.003
0.0008
0.0004
0.17
0.04
0.14
0.93
0.24
0.015
0.009
0.04
0.004
0.0003
0.38
0.34
0.004
0.13
0.01
0.24
0.78
0.0001
< 0.0001
0.75
PA = posteroanterior; QTc = corrected QT interval; RV = right ventricular.
•
4O
a5
35
35
25
25
25
~2o
~ 2o
15
15
10
10
5
5
!
Male
Female
o
5
No
Yes
Sex
40
D) p = 0 . 1 4
4o
35
30
3o
25
25
15
!
~ 20
|
is
,o
10
5
0
•
No
Yes
I
!
No
Yes
o
Histor/of MI
E) p = 0 . 2 4
4o
F) p = 0 . 9 3
35
25
15
1-
,o
5
History of VT
;! !i
History of CAD
35
~ 2o
0
C) p = 0.17
•
I
•
No
Yes
Recent Amiodarone Therapy
5
o
i!
!
!
No
Yes
Past Heart Surgery
JACC Vol. 25, No. 7
June 1995:1576-83
DEFIBRILLATION
4o
A) R = 0.04
p - 0.66
B) R = 0.33
p - 0.0008
40
•
35
4o
35
30
•
30
•
25
"20
•
• •
25
O O
20
oil,•
eleo 0 %
t
15
o,,,o
•
15
lo
8
•
•
5 "
IP.• .,o']J'~B
••••el
f~n~Oe~•
~o
Ildl~
qlo • •
O~O
•
•
10
40
o•
n
i
i
20
30
40
lo
•
5
•
go
i
u
50
60
Age (yr$)
D) R = 0.24
p = 0,015
--
i
u
u
70
80
90
80
•
oa,0°
•
120
•
i
15
~
•
i
•
i
i
t
240
i
i
40
15
•
el••
el
•
oOO
~
•
o~00
qt• ~
•
~O
5
•
i
i
2.6
I
•
30
25
10
5
•
1.6
1.8
2
2.2
2.4
Body Surface Area (m2)
C,r~ R - 0.20
p - 0.04
•
•
15
° °o "q~Jog° " °° ° "
1.4
35
20
°°°
gob
"
i
1.2
25
2o
,~:.'!
0
280
30
25
"~.
•
35
3O
,. "•- . ~ ~ . ' .
'L0
• •
160
200
Weight (Ibs)
E) R = 0.26
p = 0.009
4o
o o
•
o
go
••olin•
8
• w
I
n n
~
e'•o•
Q
O0
•
i
0
•
35
•
35
3O
25
C) R - 0 3 5
p - 0.0004
•
1579
RAITF
ET AL.
PREDICTION
THRESHOLD
1-
•alp
IOlo
~ 20
15
•
"o-oh'0
•
• o o•
•
OOo
•
"l
lO
;,," i •
0
5
°
•
Oil
°
• °
~
0o•o0•
0
o
0
4O
50
60 70
80
90 100
Resting Heart Rate (bprn)
110
120
Discussion
Several previous investigators (1-5) have reviewed their
clinical experience and found a significant but poor correlation
between a variety of clinical variables and the defibrillation
threshold or the likelihood of successful defibrillator implantation. Kelly et al. (1) reviewed the results of epicardial
defibrillator implantation in 42 of 62 patients receiving a spring
and large patch electrode configuration. A standardized defibrillation threshold protocol was followed, but the surgical
technique and waveform polarity were not controlled. Patients
taking amiodarone at the time of implantation had a higher
defibrillation threshold than that of patients who had never
taken amiodarone. Patient age, gender, cardiac diagnosis,
ejection fraction, the presence or absence of a left ventricular
aneurysm and antiarrhythmic drug therapy other than amiodarone had no significant association with the defibrillation
threshold.
Leitch et al. (3) reviewed a manufacturer's data base to
determine predictors of the defibrillation threshold in patients
undergoing epicardial defibrillator implantation. A defibrillation threshold was not determined in 137 of the 375 patients
available for study. In the remaining 238 patients the defibrillation threshold was determined but the defibrillation threshold protocol, pulsing technique, number of electrodes, electrode position and electrode polarity varied among patients.
Male gender, lower ejection fraction, higher functional class,
the use of two instead of three patches and the use of
60
80
100 120 140 160
QRS Interval (ms)
180
200
360
400
440
480
QTc Interval (ms)
520
560
Figure 2. Plots of the defibrillationthreshold (DFT) as a function of
(A) patient age, (B) bodyweight, (C) bodysurface area, (D) heart rate
at rest, (E) QRS intervaland (F) corrected QT (QTc) interval.The R
value and the p value of the linear correlation are shown, bpm beats/rain.
simultaneous rather than sequential shocks were significantly
associated with a higher defibrillation threshold on univariate
analysis. Patient age, body surface area, presence of coronary
artery disease and the use of amiodarone were not significant
predictors of the defibrillation threshold. On multivariate
analysis, patient gender, ejection fraction, and pulsing technique were independent but weak predictors of the defibrillation threshold.
More recently, Brooks et al. (4) examined the determinants
of successful nonthoracotomy defibrillator implantation (defibrillation threshold _>10 J below the maximal device output,
not the actual defibrillation threshold) in 101 patients. In their
study, defibrillators and lead systems produced by two manufacturers were utilized, and multiple lead system configurations, polarities and pulsing techniques were tested at the
discretion of the implanting physician. Significant univariate
predictors of successful implantation included smaller cardiac
diameter on chest roentgen•gram, QRS duration <120 ms,
absence of left bundle branch block, ventricular fibrillation as
the presenting arrhythmia, smaller left ventricular diastolic
diameter on echocardiogram and the absence of antiarrhyth-
1580
R A I T T ET AL.
DEFIBRILLATION T H R E S H O L D PREDICTION
o36
p = 0,0003
olBo
•
P = 0.004
35
30
•
25
~20
"
15
5
•
•
'
t
p = 0,38
35
30
30
25
25
~20
.
•
t
•
5
T
.
"~'-
;:
:,"
40
12
14
16
18
20
PA CXR Heart Width (cm)
D) R = 0,29
p = 0.004
24
22.5
40
51
27.5
32.5
37.5
42.5
47.5
...,,,.,,
.35
.4
.45
E) R = 0 . 1 0
p = 0.34
•
.5
.55
.6
.65
.7
.75
CardiothoracicRatio
40
F) R - 0.15
p - 0.13 e
•
35
35
30
•
25
• •
•
253°
25
2o
•
•
o•
15
•
O
O O
~
o°O
15
15
J•
10
_
_
j....r'rl
j•
o I o • •
•
•
0 •
..,..'2..
~20
•
ooJoooo
5
•
"':
".-
"
PA CXR ChestWidth (cm)
35
0 •
•
i
l
, '' i
7
8
9
10
11
12
Distancefrom RV Coil to Heart Border(cm)
Ooi•
ooo
0
13
O
•
o
•
O
o •
i
•
O
•o
10
•
5
•
l'":
10
o|
•00•=%0 •
.Io
5 "1 o o
•:81.*
0
"
0
22
•
30
•
8 °eee ~
•
_
:"
10
•
15
"'"
•
•
~20
15
";.;...
•OR0
•
35
....
.
JACC Vol. 25, No. 7
June 1995:1576-83
"
12
14
16
18
20
22
Distance from RV Coil to Can Electrode(cm)
Figure 3. Plots of the defibrillationthreshold (DFT) as a function of
(A) heart width on a posteroanterior chest roentgenogram (PACXR),
(B) chest width on a posteroanterior chest roentgenogram, (C) cardiothoracic ratio, (D) distance from the middle of the right ventricular
(RV) electrode to the superior heart border along the line connecting
the centers of the two electrodes, (E) distance between the right
ventricular and can electrodes and (F) horizontal distance from the
spine to the can electrode. The R value and the p value of the linear
correlation are shown.
mic drugs at the time of implantation. Patient age, height,
weight, body surface area, presence of coronary artery disease
and ejection fraction were not significantly associated with the
likelihood of successful implantation. On multivariate analysis
only smaller cardiac diameter on chest roentgen•gram and
female gender were significant independent predictors of
successful implantation. Even the most powerful predictor,
cardiac diameter, showed significant overlap between patients
with and without successful implantation (16.2 _+ 1.7 cm vs.
18.3 _+ 1.1 cm).
In these studies and in others, the investigators (1-5,18,19)
found significant correlations between certain clinical variables
and the defibrillation threshold, but the correlations were poor
and of little clinical value. They tested multiple electrode
positions, polarities, pulsing techniques and, in some cases,
different waver•tins until a configuration with an acceptable
defibrillation threshold was found. In addition, a standardized
protocol for determination of the defibrillation threshold was
4"
"
o
5
10
15
20
25
30
DistancefromSpine to Can Electrode (cm)
not always followed (2-5). It was our hypothesis that these
system-dependent sources of variability and the lack of a
uniform method to determine the defibrillation threshold
obscured the role of clinical variables in determining the
defibrillation threshold. Our findings, however, indicate that a
poor correlation between the defibrillation threshold and
clinical variables persists despite use of a uniform defibrillation
system and defibrillation threshold protocol.
Predictors of the defibrillation threshold. Our use of a
uniform defibrillation lead system, lead location, way•form,
polarity, and pulsing technique, as well as a standard protocol
to determine the defibrillation threshold, allowed identification of more clinical variables that were significantly associated
with the defibrillation threshold at a higher level of statistical
significance than that reported by past investigators. Clinical
variables significantly associated with a higher defibrillation
threshold in our patients included male gender, increased
patient height, weight and body surface area, several measures
of increased heart and chest size by chest roentgen•gram and
echocardiography, longer QTc and QRS intervals, faster heart
rate at rest and increased left ventricular mass. The results of
the multivariate analysis suggest that most of the predictive
value of these variables is related to their representation of the
same underlying property, increased heart and body size as
reflected in left ventricular mass. This correlation between left
ventricular mass and the defibrillation threshold has been
shown in one small previous study involving 11 patients (22).
The role played by heart and body size in defibrillation is
J A C C V o l . 25, N o . 7
June 1995:1576-83
DEFIBRILLATION
A) R - 0.13
p = 0.24
40
B) R - 0.03
p - 0.78
•
THRESHOLD
•
40
35
35
•
30
1581
RAITr ET AL.
PREDICTION
C) R n 0 . ~
p - 0.0001
•
35
30
25
25
•
20
•
• ,::.
~20
•
15
•
10
•
5
•
8 0 0
|
0
i
.5
40
=
•
•
•
•
•
i
i
i
J
•
n
•
•
i
n
0
.6 .7 .8 .9 1 1.1 1.2 1.3 1.4
Posterior Wall 111icknes$ (cm)
D) R = 0,45
p <0.0001
•
.4
i
.6
.8
1
u
1.2
:
1.4
u
1.6
•
• E) R = 0.25
p = 0.013
4O
•
• ..
6
7
8
9
10
F) R - 0.03
p - 0.75
•
30
25
2o
5
35
30
25
4
LV End Diastolic Diameter (cm)
35
30
i
i
1.8
$ep(al Wall Thickness (cm)
40
35
o,
/me
o• 0 0 oo
• Iljooo~i
o• • •
5
•
o ~
~,o*,,t,
lfi
j
25
"..
j,..::'.-
•
o. o d l p , O e
15
•L ~ O LO10 "n
• ~a n •
o
•~
15 ~
10
o•
•
OO •
•
15
10
•
.
5
•
0 r - , " : _ ' ,
50
150
250
350
Left Ventrictdar Mass (g)
,
450
,
550
o
understandable given the requirement that an electric field of
a certain threshold gradient (approximately 5 V/cm2) must be
applied to >90% of the myocardium to terminate ventricular
fibrillation (16). The larger the patient and the larger the heart,
the more the electric field is likely to diminish in intensity to
below this threshold at a particular pulse strength. Thus, larger
hearts and larger patients are likely to require higher delivered
energy levels to accomplish defibrillation.
Heart rate at rest was the only other independent predictor
of the defibrillation threshold: A faster heart rate at rest was
associated with a higher defibrillation threshold. The mechanism of this association is not clear, but a higher heart rate at
rest may simply be a marker for more severe heart disease that
is associated with physiologic or cellular abnormalities that
raise the defibrillation threshold.
A m i o d a r o n e . Recent amiodarone therapy was not found to
be a significant predictor of the defibrillation threshold. This
finding is unexpected because past studies (23-26) have suggested that amiodarone results in significant elevation of the
defibrillation threshold. A likely explanation is that the withholding of amiodarone for 1 to 2 weeks before testing reduced
the impact of the drug on the defibrillation threshold. Kelly et
al. (1) observed a similar phenomenon: The defibrillation
threshold of patients who had never taken amiodarone was not
different from that of patients who had stopped taking the drug
>1 month before device implantation. Patients who continued
to take amiodarone up to the time of device implantation had
significantly higher defibrillation thresholds than those of
0
.1
.2
-"
"
.3
.4
.5
Ejection Fraction
"'-"
.6
.7
o
,8'
i
35
40
t
•
•
t,, J.-,,'-
•o•0
i•
•°
• *'e* ~ f . . . .
•
• ~ Q
qP
•
o,o 000o0% •
u
=
e
,
60 65 70 75
Resistance (ohms)
80
85
45 50
i
55
e
u
e
--
4. Plots of the defibrillationthreshold (DFT) as a function of
(A) posterior wall thickness on echocardiography, (B) septal wall
thickness on echocardiography,(C) left ventricular (LV) end-diastolic
diameter on echocardiography, (I)) left ventricular mass calculated
from echocardiographic data, (E) ejection fraction and (F) shock
resistance. The R value and the p value of the linear correlation are
shown.
Figure
patients who had never taken the medication or who had
stopped taking it for >1 month. Why such an effect would
occur over a period of time significantly shorter than the
half-life of the drug is unclear. However, the finding may
reflect differences in the short- and long-term distribution of
amiodarone. Alternatively, our results may reflect the current
practice of using lower doses of the drug than had been
customary at the time of some of the earlier studies.
C l i n i c a l utility of identified p r e d i c t o r s . Despite our identification of highly significant correlations between several
clinical variables and the defibrillation threshold there is
remarkable scatter present when the defibrillation threshold of
individual patients is considered in relation to these variables
(Fig. 1 to 4). The multivariate model in fact explained only
25% of the observed variability in the defibrillation threshold.
Thus, despite controlling for all device, lead system, lead
location, waveform, polarity and defibrillation threshold measurement sources of variability in the defibrillation threshold,
we were unable to identify a set of clinical variables that could
in any reliable way predict the defibrillation threshold of an
1582
RAITT ET AL.
DEFIBRILLATION THRESHOLD PREDICTION
individual patient. These findings indicate that defibrillation is
a complex interaction of anatomy and physiology, not well
reflected in easily obtained clinical measurements, that determines the distribution of current and the excitability and the
refractoriness of cardiac membranes during ventricular fibrillation. It is possible that sophisticated methods such as finite
element analysis by means of patient-specific thoracic computed tomographic models may improve the predictability of
the defibrillation threshold in individual patients (27).
Limitations. A limitation inherent in all human studies of
defibrillation is the precision with which the defibrillation
threshold can be measured. Defibrillation is in part a statistical
phenomenon such that at any one energy level there is a
probability of defibrillation rather than a certainty of success or
failure (28). In animal studies the defibrillation threshold is
often defined as the energy level with a 50% likelihood of
defibrillation, in human studies it is neither practical nor safe
to perform a large enough number of defibrillations so that this
energy level can be precisely determined. Thus, in any one
patient chance may lead to significant under- or overestimation
of the defibrillation threshold. Our defibrillation threshold
measurement protocol minimized this effect by using an initial
energy level near the expected mean defibrillation threshold of
the entire study group. In addition, the energy step-skipping
protocol allows relatively small increments in energy to be
tested near the defibrillation threshold while minimizing the
overall number of inductions of ventricular fibrillation. Nevertheless, it is possible that the inherent variability in the
defibrillation threshold measurement obtained with this
method and other methods reasonable in human studies and
clinical practice may account for a significant proportion of the
75% of the defibrillation threshold not accounted for by
clinical variables. Nevertheless, it is the results of this type of
defibrillation threshold testing that physicians would like to be
able to predict.
A second potential limitation of this study is the concern
that the result may be relevant only to the specific lead
configuration, pulsing method and waveform tested. This
limitation is, of course, inherent in any study that controls
these factors. This issue is made less important because our
results suggest that there is not an inherent determination of
the defibrillation threshold by easily measured clinical factors
but, instead, a significant but minimal influence. Testing a
different waveform or electrode configuration may increase or
decrease the modest influence of individual clinical variables,
but it is unlikely to change the basic finding that something
other than these clinical measures accounts for the preponderance of the observed variability in the defibrillation threshold
in clinical practice.
Conclusions. The defibrillation threshold cannot be reliably predicted for individual patients from standard clinical
variables. Though certain clinical variables are associated with
a higher or lower defibrillation threshold, the predictive value
is low. We do not advocate withholding an attempt at transvenous defibrillator implantation on the basis of a clinical
variable such as increased left ventricular mass or patient
JACC Vol. 25, No. 7
June 1995:1576-83
weight that tends to be associated with an increased defibrillation threshold. The defibrillation threshold is probably a
function of a complex interaction of anatomic, physiologic and
cellular variables that are not adequately represented by easily
obtainable clinical information.
We thank Charles Troutman, RN and Jill Anderson, RN for nursing care of the
patients, Joan McDaniel for secretarial assistance and Susan Shattuc, MS for
preparation of the figures.
References
1. Kelly PA, Cannom DS, Garan H, et al. The automatic cardioverterdefibrillator: efficacy, complications and survival in patients with malignant
ventricular arrhythmias. J Am Coil Cardiol 1988;11:1276-86.
2. Epstein AE, Ellenbogen KA, Kirk KA, Kay GN, Dailey SM, Plumb VJ, and
the High Defibrillation Threshold Investigators. Clinical characteristics and
outcome of patients with high defibrillation thresholds: a multicenter study.
Circulation 1992;86:1206-16.
3. Leitch JW, Yee R, and the Multicenter Pacemaker-CardioverterDefibrillator (PCD) Investigators Group. Predictors of defibrillation efficacy
in patients undergoing epicardial defibrillator implantation. J Am Coll
Cardiol 1993;21:1632-7.
4. Brooks R, Torchiana D, Vlahakes G, Ruskin J, McGovern B, Garan H.
Successful implantation of cardioverter-defibrillator systems in patients with
elevated defibrillation thresholds. J Am Coil Cardiol 1993;22:569-74.
5. Brooks R, Garan H, Torchiana D, et al. Determinants of successful
nonthoracotomy cardioverter-defibrillator implantation: experience in 101
patients using two different lead systems. J Am Coll Cardiol 1993;22:183542.
6. Ideker RE, Wolf PD, Afferness C, Krassowska W, Smith WM. Current
concepts for selecting the location, size, and shape of defibrillation electrodes. PACE 1991;14:227-40.
7. Swartz JF, Karasik P, Fletcher RD. Sub-scapular patch position enhances
multiple pathway non-thoracotomy defibrillation in humans [abstract]. Circulation 1992;86 Suppl I:I-441.
8. Bhandari AK. lntraoperative defibrillating efficacy of endocardial subcutaneous lead configurations [abstract]. Circulation 1992;86 Suppl I:I-790.
9. Jung W, Pfeiffer D, Moosdorf R, et al. Effects of transvenous electrode
configuration and shock waveforms on defibrillation threshold in man
[abstract]. Circulation 1992;86 Suppl I:I-792.
10. Bardy GH, Troutman C, Johnson G, et al. Electrode system influence on
biphasic waveform defibrillation efficacy in humans. Circulation 1991;84:
665-71.
11. Bardy GH, Ivey TD, Allen MD, Johnson G, Greene HL. Evaluation of
electrode polarity on defibrillation efficacy. Am J Cardiol 1989;63:433-7.
12. O'Neill PG, Boahene KA, Lawrie GM, Harvill LF, Pacifico A. The automatic implantable cardioverter-defibrillator: effect of patch polarity on
defibrillation threshold. J Am Coil Cardiol 1991;17:70%11.
13. Bardy GH, Ivey TD, Allen MD, Johnson G, Greene HL. Prospective
comparison of sequential pulse and single pulse defibrillation with use of two
different clinically available systems. J Am Coil Cardiol 1989;14:165-71.
14. Hsia HH, Kleiman N, Flores BT, Marchlinski FE. Comparison of simultaneous versus sequential pathways for defibrillation energy delivery [abstract].
Circulation 1992;86 Suppl I:I-790.
15. Bardy GH, lvey TD, Allen MD, Johnson G, Mehra R, Greene HL. A
prospective randomized evaluation of biphasic versus monophasic pulses on
defibrillation efficacy in humans. J Am Coll Cardiol 1989;14:728-33.
16. Mehra R, DeGroot PJ, Norenberg MS. Energy waveforms and lead systems
for implantable defibrillators. In: Lfideritz B, Saksena S, editors. Interventional Electrophysiology. Mount Kisco (NY): Futura, 1991:377-94.
17. Feeser SA, Tang AS, Kavanagh KM, et al. Strength-duration and probability
of success curves for defibrillation with biphasic waveforms. Circulation
1990;82:2128-41.
18. Bardy GH, Troutman C, Poole JE, et al. Clinical experience with a
tiered-therapy, multiprogrammable antiarrhythmia device. Circulation 1992;
85:1689-98.
JACC Vol. 25, No. 7
June 1995:1576-83
19. Bardy GH, Hofer B, Johnson G, et al. Implantable transvenous cardioverterdefibrillators. Circulation 1993;87:1152-68.
20. Bardy GH, Johnson G, Poole JE, et al. A simplified, single-lead unipolar
transvenous cardioversion-defibrillation system. Circulation 1993;88:
543-7.
21. Devereux RB, Alonso DR, Lutas EM, et at. Echocardiographic assessment
of left ventricular hypertrophy: comparison with necropsy findings. Am J
Cardiol 1986;57:450-8.
22. Chapman PD, Sagar KB, Wetherbee JN, Troup PJ. Relationship of left
ventricular mass to defibrillation threshold for the implantable defibrillator:
a combined clinical and animal study. Am Heart J 1987;114:274-8.
23. Frame LH. The effect of chronic oral and acute intravenous amiodarone
administration on ventricular defibrillation thresholdusing implanted electrodes in dogs. PACE 1989;12:339-46.
24. Huang SK, Tan de Guzman WL, Chenarides JG, Okike NO, Vander Salm
RAITT ET AL.
DEFIBRILLATION THRESHOLD PREDICTION
25.
26.
27.
28.
1583
TJ. Effects of long-term amiodarone therapy on the defibrillation threshold
and the rate of shocks of the implantable cardoverter-defibrillator. Am
Heart J 1991;122:720-7.
Jung W, Manz M, Luderitz B. Effects of antiarrhythmic drugs on defibrillation threshold in patients with the implantable cardioverter defibrillator.
PACE 1992;15:645-8.
Jung W, Manz M, Pizzulli L, Pfeiffer D, Luderitz B. Effects of chronic
amiodarone therapy on defibrillation threshold. Am J Cardiol 1992;70:
1023-7.
Karlon WJ, Eisenberg SR, Lehr JL. Effects of paddle placement and size on
defibrillation current distribution: a three-dimensional finite element model.
IEEE Trans Biomed Eng 1993;40:246-55.
McDaniel WC, Schuder JC. The cardiac ventricular defibrillation threshold:
inherent limitations in its application and interpretation. Med Instrum
1987;21:170-6.