1510

Cardiac Potential and Potential Gradient Fields

Generated by Single, Combined, and Sequential
Shocks During Ventricular Defibrillation
J. Marcus Wharton, MD; Patrick D. Wolf, MS; William M. Smith, PhD; Peng-Sheng Chen, MD;
David W. Frazier, MD; Seitaro Yabe, MD; Ned Danieley, MS; and Raymond E. Ideker, MD, PhD

Background. Potential gradient field determination may be a helpful means of describing the effects of
defibrillation shocks; however, potential gradient field requirements for defibnrllation with different
electrode configurations have not been established.
Methods and Results. To evaluate the field requirements for defibrillation, potential fields during
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017

defibrillation shocks and the following ventricular activations were recorded with 74 epicardial electrodes
in 12 open-chest dogs with the use of a computerized mapping system. Shock electrodes (2.64 cm2) were
attached to the lateral right atrium (R), lateral left ventricular base (L), and left ventricular apex (V).
Four electrode configurations were tested: single shocks of 14-msec duration given to two single
anode-single cathode configurations, R: V and L: V, and to one dual anode-single cathode configuration,
(R+L):V; and sequential 7-msec shocks separated by 1 msec given to R:V and L:V (R:V-*L:V).
Defibrillation threshold (DFT) current was significantly lower for R:V--L:V than for the other
configurations and markedly higher for L:V. Despite these differences, the minimum potential gradients
measured at DEFI were not significantly different (approximately 6-7 V/cm for each electrode configura-
tion). Potential gradient fields generated by the electrode configurations were markedly uneven, with a
15-27-fold change from lowest to highest gradient, with the greatest decrease in gradient occurring near
the shock electrodes. Although gradient fields varied with the electrode configuration, all configurations
produced weak fields along the right ventricular base. Early sites of epicardial activation after all
unsuccessful shocks occurred in areas in which the field was weak; 87% occurred at sites with gradients
less than 15 V/cm. Ventricular tachycardia originating in high gradient areas near shock electrodes
followed 11 of 67 successful shocks.
Conclusions. These data suggest that 1) defibrillation fields created by small epicardial electrodes are
very uneven; 2) achievement of a certain minimum potential gradient over both ventricles is necessary for
ventricular defibrillation; 3) the difference in shock strengths required to achieve this minimum gradient
over both ventricles may explain the differences in DIFTs for various electrode configurations; and 4) high
gradient areas in the uneven fields can induce ectopic activation after successful shocks. (Circuaion
1992;85:1510-1523)
KEY WORDS * defibrillation * fields, gradient

T odefibrillate with the least amount of energy, the
potential field generated by the defibrillating
shock must be optimized. The potential gradi-
ent has been reported to be closely related to the
success or failure of defibrillation.' If a certain mini-
mum potential gradient is necessary over the myocar-
dium to achieve defibrillation, shocks that do not pro-
duce this minimum potential gradient will not be
successful. Furthermore, the defibrillation threshold
(DFT) should be determined by the shock strength
required to obtain this minimum gradient across the
myocardium; thus, potential gradients may explain the
relative efficacy of different electrode configurations.
Though a few studies have measured potential gradients
and gradient fields during electrical shocks in sinus
rhythm or ventricular fibrillation,2-5 systematic evalua-
tion of field requirements for multiple electrode config-
urations to determine the minimum magnitude and
spatial requirements of potential gradient fields has not
been performed. In addition, mapping of potential
gradient fields may help to explain the increased efficacy
of some sequential shock electrode configurations. The

All editorial decisions for this article, including selection of
reviewers and the final decision, were made by a guest editor. This
procedure applies to all manuscripts with authors from the Uni-
versity of California San Diego or UCSD Medical Center.
From the Departments of Medicine and Pathology (J.M.W.,
P.D.W., W.M.S., D.W.F., S.Y., N.D., R.E.I.), Duke University Med-
ical Center, Durham, N.C., and the Cardiology Division (P.-S.C.),
University of California Medical Center, San Diego, Calif.
Supported in part by National Institutes of Health research
grants HL-17670, HL-28429, HL-33637, HL-07063, HL-44066,
and HL-42760; American Heart Association North Carolina Af-
filiate grant-in-aid 1985-86 A-06; and a grant from the Fannie E.
Ripple Foundation. J.M.W. received the North American Society
of Pacing and Electrophysiology Young Investigator Award for a
presentation of a preliminary version of this article.
Address for correspondence: J. Marcus Wharton, MD, PO Box
3816, Duke University Medical Center, Durham, NC 27710.
Received January 22, 1991; revision accepted November 21, 1991.

present study was designed to 1) describe the potential
and potential gradient fields generated by electrode
configurations with different defibrillation efficacies; 2)
correlate the field characteristics with the efficacy of
each electrode configuration; 3) describe the field char-
acteristics necessary for defibrillation; and 4) define
further the relation of field strength to the site of initial
ventricular activation after successful and unsuccessful
defibrillation shocks.
Methods
Surgical Preparation
In 12 mongrel dogs (mean weight, 19.74.7 kg),
anesthesia was induced with intravenous pentobarbital
30-35 mg/kg and maintained with a dose of 0.05
mg/kg/min.6 Skeletal muscle paralysis was obtained
with intravenous succinylcholine 1 mg/kg and main-
tained with a dose of 0.25-0.50 mg/kg/hr as needed to
prevent skeletal muscle contraction.7 The dogs were
intubated and mechanically ventilated (model 607, Har-
vard Apparatus Company, Inc.). A midline sternotomy
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
was performed, and the heart was suspended in a
pericardial cradle. The nylon sock, which was placed
over the ventricles for mapping purposes (see below),
helped prevent desiccation of the exposed portions of
the heart; warm saline was also applied repeatedly to
the sock to keep it moist. Warm lactated Ringer's
solution was infused continuously through a peripheral
intravenous cannula, and blood pressure was monitored
through a femoral artery cannula connected to a trans-
ducer (model 1280, Hewlett-Packard Co.). Arterial
blood was sampled every 30-60 minutes (or more
frequently if indicated) and pH, Pco2, Po2, HCO2-, Na+,
K+, and Ca24 levels were monitored and corrected as
necessary to maintain these parameters within normal
limits. Core temperature was monitored with a tongue
thermistor and maintained at 37C using a heating
blanket. A heating lamp warmed the open chest. All
studies were performed in accordance with the guide-
lines for use of experimental animals established by the
American Physiological Society.
Recording and Shock Electrodes
A nylon sock with 72 evenly spaced button electrodes8
was applied over the heart and secured to the atria,
atrioventricular groove fat, or pericardium to ensure
stability of location. The button electrodes were placed
in seven rows from apex to base (Figure 1A), with the
basal row extending to the atrioventricular groove. Each
button encases two electrodes with centers 1.5 mm apart
(1.0 mm edge to edge). Unipolar or bipolar recordings
were obtained from these electrodes depending on the
setting of the mapping system (see below). A clamp
electrode attached to the left leg served as the reference
for unipolar recordings.
Three solid circular titanium mesh electrodes with a
diameter of 2.0 cm (total surface area, 3.14 cm2) and
with a bipolar button recording electrode (surface area
of plastic button, 0.50 cm2) centered underneath were
used to deliver shocks (Figures 1A and 1B). The total
area of the shock electrode exposed to myocardium and
not bounded by insulator was a circular ring with a
width of 0.6 cm (surface area, 2.64 cm2). Shock elec-
trodes were sewn securely to the epicardium of the
Wharton et al Defibrillation Gradients 1511
lateral wall of the right atrium near the junction of the
superior vena cava (electrode R in Figures 1A and 1B),
the lateral free wall of the left ventricle adjacent to the
atrioventricular groove (electrode L), and the left ven-
tricular apex (electrode V). The location of the L and V
electrodes was varied slightly to avoid placement over or
near coronary arteries. Two pairs of stainless steel wires
were inserted into the right ventricular free wall approx-
imately 1 cm apart for ventricular pacing and sensing.
Defibrillation Protocol
After induction of ventricular fibrillation with a short
burst of 60-Hz alternating current in the open-chest
preparation, single truncated exponential defibrillation
shocks with a duration of 14 msec were given with a
specially designed 750-microfarad defibrillator (Inter-
medics, Inc.) through one of three anode-cathode
electrode configurations, abbreviated R:V, L:V, and
(R+L):V. The last abbreviation indicates that the R
and L electrodes are combined as a common anode.
One sequential shock (abbreviated R:V--L:V) was
tested in which a 7-msec R: V was followed by a 7-msec
L: V separated by a pause of 1 msec.9 The 14-msec
monophasic waveform had a 16% tilt when delivered
across a test resistor of 100 Ql.
DFTs were determined using a modified Purdue
method.10 The initial shocks were given at an estimated
strength of 4 J (2 J to each shock for the sequential pair)
to each of the electrode configurations in random order.
The shock energy to be delivered was estimated by
measuring the current during a 100-V shock delivered
in sinus rhythm from an electrode configuration, calcu-
lating the impedance, and then calculating the voltage
necessary to deliver 4 J (or 2 J sequentially) assuming no
change in impedance. Because impedances for R: V and
L: V were different, potentials required to deliver 2 J to
each component of the sequential shock were not the
same. Each shock was given approximately 15 seconds
after the induction of ventricular fibrillation; when it
was unsuccessful, a 10-30-J rescue shock (defibrillator
model 7802B, Hewlett-Packard) was given through
hand-held paddles placed on the pericardium over the
right and left ventricles. Temporary pacing and/or brief
cardiac massage was given during periods of asystole
after cardioversion. Another shock was not attempted
for at least 5-10 minutes or until baseline heart rate and
blood pressure were reattained.6,11,12
When a defibrillation attempt failed, the energy level
for the subsequent shock was increased by 20% above the
delivered energy from the preceding unsuccessful shock
for that electrode configuration, and defibrillation was
reattempted. Estimation of the voltage required to deliver
the desired energy for each electrode configuration was
based on the impedance measured from the preceding
shock with the same electrode configuration. When a
defibrillation shock was successful, the energy for the next
shock for that electrode configuration was decreased by
10% until defibrillation was not obtained. Each electrode
configuration was tested in random order for each incre-
mental change in energy. The lowest energy resulting in
defibrillation was called the DFT.
Electrode Recordings
Before and after each attempted defibrillation, epi-
cardial activation was recorded by the 75 bipolar button
 1512 Circulation Vol 85, No 4 April 1992
A. . ^
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
C.
RVOT
APEX ' I 20V
lOOmnsec II..
loomw ~~~~~~~~11
I>VIII
electrodes (74 of which were on the ventricles). Ampli-
fier gains were set for each channel for optimal signal
recording. Electrogram signals were filtered to pass
0.1-500 Hz before and after attempted defibrillation.13
Ten milliseconds before a shock, recording was switched
to unipolar at a second set of gains (previously deter-
mined to be appropriate for the voltage of the shock to
be delivered) with a low-pass filter of 500 Hz and
high-pass filter direct-current coupled14 (Figure 1C).
Ten milliseconds allowed adequate time to establish a
baseline for measuring potentials generated by the
shocks.14 A voltage attenuator produced by a 1-Gfl
FIGURE 1. Schematic representation of an-
terior (panel A) and right lateral (panel B)
views of dog heart illustrating locations ofthe
three shock electrodes (large cross-hatched or
unfilled circles labeled L, V and R). The right
atrial shock electrode (R) in the anterior view
(panel A) and the left lateral free wall shock
electrode (L) in the right lateral view (panel
B) are not cross-hatched to indicate that they
are located behind the surfaces illustrated.
Bipolar recording electrode within an insu-
lated button is shown in the center of each
shock electrode. Smallfilled circles represent-
ing locations of recording electrodes across
the ventricles are shown in the anterior view
(panelA) only. In the right lateral view (panel
B), an imaginary plane formed by the three
shock electrodes is outlined by straight lines
between the electrodes. Note that right ven-
tricular outflow tract is farther away from this
imaginary plane than is the inferiorportion of
the ventricles. Panel C: Simultaneous record-
ing of an unsuccessful 252-V (R+L): V (dual
anode-single cathode configuration) shock
displaying recordings from two electrodes,
one on the right ventricular outflow tract
I lOmV (RVOT) and the other on the left ventricular
apex. Solid bar at the top offigure indicates
25-msec period that the mapping system is
recording unipolar with a set of gains and
filter settings appropriate for the shock and
with voltage attenuator switched on. Four
intervals can be identified: A, sharp initial
deflection as the voltage attenuator is
switched on; B, period offlat baseline lasting
approximately 6 or 7 msec from which the
magnitude of the shock potentials is refer-
enced; C, abrupt square deflection (with
spiked leading and trailing edges caused by
stray capacitance) representing the 14-msec
truncated exponential shock; and D, another
abrupt, high-amplitude deflection after
switching off voltage attenuators. Both before
and after the shock period, the system is
recording bipolar with a different set of gains
and filters. Vertical scales on right side of
figure indicate 10 mVand 20 Vduring bipolar
electrogram and unipolar shock recording,
respectively.
resistor at the front end of the amplifier was simultane-
ously used with the other changes to reduce the elec-
trode potential by a factor of 1,000. During the shock,
epicardial potentials were measured by the 75 button
electrodes, which included the three electrodes located
beneath the shock electrodes. The potential delivered to
the shock electrodes was also directly measured. Cur-
rent delivered to each electrode pair for single, com-
bined, and sequential electrode configurations was mea-
sured on an oscilloscope. The voltage attenuator was
switched off 1 msec after the shock, and recording of
bipolar electrograms was resumed (Figure 1C).

TABLE 1. Global and Field Parameters at Defibrillation Threshold
R:V
Global parameters
Voltage (V) 272.733.6
Current (A) 2.24+0.54*
Energy (J) 8.72.8*
Field parameters
Minimum gradient (V/cm) 6.71.1
Maximum gradient (V/cm) 129.388.1
R:V and L:V, single anode-single cathode configurations; (R+L):V, dual anode-single cathode
R:V-*L:V, sequential 7-msec shocks separated by 1 msec given to R:V and L:V.
*p<0.05 with respect to R:V--L:V.
tpO0.005 with respect to R:V->L:V.
:Represents field parameter from maximum combined gradient field (see text).
After completion of the study, the dog was killed by
induction of ventricular fibrillation. The button and shock
electrode locations were marked with color-coded pins.
Their locations were subsequently transcribed to a com-
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
puter-generated, three-dimensional display of the dog
heart for generation of surface maps.
Data Analysis
Data were digitized at 1,000 Hz and recorded on
videocassette for subsequent analysis.15,16 Shock poten-
tials at the median time of the shock were determined
for every recording electrode by a computer program
with known calibration signals. Total delivered potential
was calculated from the measured shock electrode
potentials as the anodal potential or mean anodal
potential for (R+L):V, minus the cathodal potential.
Total energy delivered was calculated for all dogs
from the equation
J=VIt
where J is energy in joules, V is mean delivered
potential in volts, I is mean current in amperes, and t is
the duration of the shock in seconds. In five dogs, a
digital oscilloscope was used to store the measured
current and potential waveforms, and the power curve
was integrated over time to obtain the delivered energy.
Energy measurements by this means were nearly iden-
tical to those made with the above equation (r=0.9998;
slope, 0.99).
Impedance was calculated from Ohm's law, assuming
a primarily resistive element to impedance in the range
of potentials used in this study (see "Appendix"):
R= AE/I
For total impedance, AE equals the total delivered
potential. For electrode-myocardial interface imped-
ance, AE is the potential difference from shock elec-
trode to the button electrode underneath it. Cardiac
impedance was calculated as the difference between
total and electrode-myocardial interface impedance.
Because intracardiac and shunt current were not mea-
sured in the parallel circuit of (R+L):V, direct mea-
surement of the associated cardiac impedance was not
possible; it was estimated by averaging the potentials
measured underneath the anodal shock electrodes and
assuming a single circuit impedance.
Wharton et al Defibrillation Gradients 1513
L:V (R+L):V R:V--L:V
430.2102.9t 256.6+64.7 238.745.6
4.101.47t 2.700.92* 1.900.36
26.417.7t 10.26.4 6.42.2
7.42.0 5.81.3 6.61.81
188.892.4t 136.9103.4 109.462.2t
configuration;
Potential gradients were calculated as described pre-
viously.4 Briefly, the nearest neighbors of each recording
electrode were determined, and the interelectrode dis-
tance of each electrode to its neighbors was calculated
from their x-y-z coordinates. From the potential differ-
ences and interelectrode distances between an elec-
trode and each of its nearest neighbors, a method that
minimized the summed squared error was used to
estimate the potential gradient at the recording elec-
trode. The minimum and maximum gradient for each
defibrillation electrode configuration for each dog was
determined from the lowest and highest gradient, re-
spectively, from any recording electrode for a given
shock.
Postshock epicardial activation times at each record-
ing electrode, referenced to the onset of the defibrilla-
tion shock, were determined by using the peak of
monophasic waveforms or the maximum potential
change of biphasic or multiphasic waveforms as the
point of activation. The earliest site of postshock ven-
tricular activation was defined as the recording elec-
trode site that first recorded activation after the shock.
Other early sites were defined as electrode sites with
activation times earlier than surrounding neighbor elec-
trodes. Epicardial activation times, potentials, and po-
tential gradients were projected onto two-dimensional
representations of the dog heart for visual analysis.
Data are expressed as meanSD. Repeated-mea-
sures analysis of variance17 was performed to compare
measured and calculated values between the multiple
electrode configurations. Pearson correlation coeffi-
cients were used to compare different means of energy
calculation and to compare potential fields to test the
rule of superposition. A probability value less than 0.05
was considered significant.
Results
Total Voltage, Current, and Energy Required at
Defibrillation Threshold
Of the 12 dogs studied, eight had their lowest DFT in
terms of energy with R: V->L: V, two with R: V, and
two with the (R+L):V electrode configuration. The
sequential shock configuration required significantly
lower current (p<0.05) for defibrillation than any other
configuration and tended to require less energy: p= 0.07
compared with (R+L):V and pc0.05 compared with
R: V and L: V (Table 1). Voltage requirements for
defibrillation with the sequential shock were not signif-
 1514 Circulation Vol 85, No 4 April 1992
icantly different from R:V and (R+L):V. R:V and
(R+L): V were similar in terms of energy and voltage,
but (R+L): V required significantly more current than
R:V (p<0.05). On the other hand, L:V was markedly
worse in all parameters. At DFT, mean voltage and
current for L:V were both approximately 1.5 times
higher and energy was approximately three times higher
than for the other electrode configurations (pc0.005
for all parameters).
There were 329 shocks delivered: 81 R:V, 114 L:V,
80 (R+L):V, and 54 R:V->L:V. The mean numbers of
shocks necessary to obtain DFT for each of these
electrode configurations were 72, 102, 73, and
51, respectively. Determination of the DFT for L:V
required a significantly greater number of shocks than
for the other electrode configurations.
Potential and Potential Gradient Fields
The potential fields, potential gradient fields, and the
postshock activation sequences were mapped for all
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
unsuccessful and successful shocks. For each electrode
configuration, the potential and potential gradient fields
generated were similar among different experimental
animals. A representative example of maps of the
potential field and potential gradient field for each
component of the sequential shock configuration is
shown in Figure 2. Each component shock of
R:V->L:V was similar to R:V and L:V shocks of
equivalent magnitude given singularly. Potential mea-
sured at the recording electrode immediately under-
neath the shock electrode was approximately two thirds
of the measured delivered potential for both compo-
nents of the sequential shock. Mean potential decrease
at the electrode-myocardial interface ranged from 27%
to 29% of the applied potential for all of the electrode
configurations tested. Decline in potential across the
myocardium was relatively symmetrical with respect to
the interelectrode axis of each configuration and was
most rapid near the shock electrodes. As can be seen in
Figures 2A and 2B, potential change was small across
the base of the right ventricle for both R: V and L: V.
Potential change was also small across the entire base of
the left ventricle with R: V shocks and across the
posterior left ventricular base with L: V shocks.
When the combined anodal configuration (R+L):V
was used, the potential field was similar to R: V except
for a rapidly decreasing component which rapidly de-
creased with distance from the left ventricular anode
(Figure 3A). From the rule of superposition, the poten-
tial field generated by (R+L): V should be equal to the
sum of the potential fields generated by R: V and L: V
if current flow were equal between each component of
(R+L): V and the corresponding R: V and L: V shocks.
Equal currents (0.1 A) among all necessary compo-
nent shocks were available for four shock strengths in
two dogs. The correlation between the actually mea-
sured (R+L): V potential field and the field calculated
from superpositioning equicurrent R: V and L: V fields
for these four shocks was excellent (r=0.993-0.998).
The small size of the shock electrodes resulted in very
inhomogeneous potential gradient fields (Figures 2C
and 2D). The unevenness of the potential gradient fields
for each electrode configuration is quantitated by com-
parison of minimum and maximum gradients at DFT in
Table 1. There was a 19.5-+f13.2-, 27.6 + 17.7-,
23.3 15.3-, and 15.7+4.4-fold change between the min-
imum and maximum measured potential gradient at the
DFT for R: V, L: V, (R+L): V, and the composite field
(see below) of (R: V)->(L: V), respectively. Paralleling
the changes seen in the potential fields, potential gra-
dient decline was most rapid near the shock electrodes
and more gradual at sites more distant from the shock
electrodes. Relatively weak potential gradient fields
(less than 10 V/cm) were produced across the base of
the right ventricle by all electrode configurations used.
R: V also generated weak fields across most of the base
of the left ventricle and L: V across the posterior base of
the left ventricle. For (R+L): V, the gradient field was
less than 10 V/cm across the base of both ventricles
except for a relatively small area surrounding the L
electrode (Figure 3B). Except for the area enhanced by
the L anode, the gradient field was not greatly different
from that of R: V, which may explain their similar
DFTs. Figure 4 graphically illustrates the distribution of
low gradient areas for each electrode configuration.
As can be anticipated from the individual gradient
fields for R: V and L: V, neither component of the
sequential shocks obtained high gradients across the
right ventricular base or across portions of the base of
the left ventricle anteriorly and posteriorly near the
right ventricle and distant from the L electrode. To
illustrate this, composite maps were generated by using
the maximum gradient measured from either of the two
component shocks at each electrode site (Figure 2E).
This is justifiable because both 7- and 14-msec trun-
cated exponential shocks occur on the relatively flat
portion of the strength-duration curve10 and presum-
ably should have approximately the same potential
gradient field requirements for defibrillation. Such com-
posite maps demonstrated that neither of the sequential
shocks created potential gradients greater than 10 V/cm
across the base of the right ventricle. In some experi-
ments, gradients greater than 10 V/cm were also not
created by the sequential shocks across the anterior
and/or posterior base of the left ventricle. However,
52+19% of recording electrodes with gradients less
than or equal to 10 V/cm during the R: V component of
the sequential shock at DFT recorded gradients greater
than 10 V/cm during the L:V component. The mini-
mum gradient for each component of the sequential
shock at DFT was 5.3+0.8 V/cm and 4.40.9 V/cm,
respectively. These were significantly less (p=0.012)
than the minimum gradient of 6.61.8 V/cm for the
composite field. The ability of two sequential fields of
lower minimum gradient to generate a composite field
with higher minimum gradient presumably accounts for
the greater efficacy of the sequential shock compared
with either component alone. The minimum gradients
at DFT for the R: V and L: V components of the
sequential shock were significantly less (p<0.01) than
for those at DFI' for the respective single 14-msec
shocks. However, the minimum gradients at DFT for
the composite field of the sequential shock and for the
fields of the single R: V and L: V shocks were similar.
Minimum Gradient and Defibrillation Parameters
R:VL:V required significantly less and L: V re-
quired significantly more current to defibrillate than
 Wharton et al Defibrillation Gradients 1515

A B
R:V -.L:V R:V -* L: V
+27
3
-II
6 -5 i`ij
10

-ill 14
a -B
24 1 - 25
7--0 -

_e X7/ -'37~
~ ~ ~ ~ ~

RI5-81
POTENTIAL FIELD POTENTIAL FIELD

2? 29 26 26
C S 467 -
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017

GRADIENT FIELD GRADIENT FIELD

E
18
41T
Io
1~ ~ ~ ~ ~ ~
1~~~~~~~~~~~~~~~~~~~~I

II

'43~ ~~lr

MAXIMUM COMBINED GRADIENT FIELD POST-SHOCK ACTIVATION

FiGURE 2. Maps ofpotentialfieldsforsequential 208-VR:V(two single anode-single cathode configuration) (panel A) and 205-V
L: V (two single anode-single cathode configuration) (panel B) shocks are iUlustrated for an unsuccessful defibriUation attempt.
Left-hand figure of each panel represents anterior half and right-hand the posterior half ofthe ventricles. Numbers indicate location of
recording epicardial electrodes andpotential in volts (relative to the left leg) recorded at each site. Boxed-in numbers represent measured
potential delivered to shock electrodes: number at upper left-hand corner in panelA represents the anode at the right atrium (R); number
at left base in panel B represents the left basal anode (L); nuwberat left-hand apex ofbothpanelsA and B represent the left ventricular
apical cathode (V). Numbers immediately beneath the right atrial or to the left ofthe left basal anodalpotentials and above the cathodal
potentials indicate potentials measured by electrode immediately beneath shock electrodes. Isopotential lines are separated by 20 V Note
relative symmetry of isopotential lines; they approximately describe circles perpendicular to the long axis of the heart during the R: V
shock. During the L: V shock, there is also relative symmetry of the potential field along the axis between L and V Fields generated
between each 7-msec duration component of the sequential shock were similar to the corresponding 14-msec single shock of similar
deliveredpotentiaL Panels C and D: DerivedpotentialgradientfieldsforR: Vand L: Vcomponents ofsequential shock illustrated above
are shown. Numbers represent potential gradient at each site in V/cm. Lowest gradients ('10 V/cm) are enclosed in cross-hatched area.
Isogradient lines are separated by 20 V/cm and parallel the symmetry ofpotential fields. During R: V shock (panel C), gradient at
electrode beneath apical cathode was 48 V/cm. During the L: Vshock (panel D), gradient at electrode underneath left ventricular basal
anode was 48 Vlcm and underneath apical cathode was 43 V/cm. Panel E: To observe the effect of sequential shocks on improving
gradient distribution across the epicardium, maximum gradient from either shock at each electrode site was displayed. "Maximum
combined gradientfield" for sequential shock illustrated in panels C and D is shown in panel E. Note that extent of myocardium not
obtaining a gradient greater than 10 V/cm is substantially reduced compared with gradientfields for component shocks. Panel F: Map
ofepicardial activation after unsuccessful sequential shock. Numbers represent times in milliseconds when epicardial activation occurs
at the recording electrode, with time zero being the start of the shock. Filled circles represent sites of electrodes in which adequate biolar
recording of activation was not obtained. Isochronal lines are separated by 20 msec. Epicardial activation map shows three sites ofearly
epicardial activation arising along the atrioventricular groove in which neither shock electrode generated a strong field. Thick arrow
indicates site of earliest postshock epicardial activation; thin arrows indicate other somewhat later sites of epicardial activation.
Subsequently, there is rather uniform spread across the ventricles from these foci.
 1516 Circulation Vol 85, No 4 April 1992
A R: V or (R+L): V (Table 1). In addition, the potential
(R+ l) :V
+18
RV LV LV RV
X
-~I
1 20-1-4
FIELD
B energy required for defibrillation, there was not a
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
GRADIENT FIELD
1211
04 1~~~~~~
POST-SHOCK ACTIVATION
FIGURE 3. Maps. Panel A: Potentialfieldfor 220-Vsubthresh-
old shock with the (R+L):V (dual anode-single cathode)
electrode configuration from the same experimental animal as
in Figure 2 Potential delivered to right atrial anode was +35 V
(upper left-hand corner offigure), to the left ventricular anode,
+36 V, and to the apical cathode, -184 V Potentials measured
underneath respective shock electrodes were +18, +21, and
-124 V Potential drop across anodes was similar. Potential
drops rapidly around the L shock electrode with little distortion
otherwise of the field generated by R: V (single anode-single
cathode configuration) (Figure 2A). Isopotential lines are thus
relatively symmetrical across apical half of myocardium. Panel
B: Gradientfieldforthe same (R+L): Vshock. Largest gradient
(82 Vkcm) is located beneath apical cathode. Gradient measured
underneath left ventricular anode was 34 V/cm. Gradientfield is
weak across most of the base of the ventricles except for area
immediately surrounding the L electrode. Panel C: Map of
epicardial activity after shock illustrated in panel A reveals two
electrodes recording the same postshock activation time of 63
msec, presunably resultingfrom an area of earlier activation at
a site between the two recording electrodes (thick arrow) prop-
agating out to recording electrodes at approximately the same
time. Two other earl sites (thin arrows) ofpostshock activation
are seen, one at the base of the left ventricle and the other along
the right ventricular free wall both in areas in which gradient
field is weak.
fields generated were different for each electrode con-
figuration. Despite these differences, the minimum po-
tential gradients measured at DFT for each of these
electrode configurations were approximately 6-7 V/cm
and were not significantly different (Table 1). This
relation is also true for the composite field of the
sequential shock configuration; however, the individual
component shock fields had minimum gradients that
were somewhat lower (see above). The mean maximum
gradient for L:V was significantly larger than for the
other electrode configurations (p<O.OO9), although the
latter were not different from each other (Table 1).
Thus, despite the differences in electrode configuration,
in generated potential fields, and in voltage, current and
significant difference in the minimum potential gradient
generated by any of the electrode configurations at their
DFTs. This implies that there is a minimum potential
gradient that must be obtained over both ventricles for
electrical defibrillation in normal myocardium. For se-
quential shocks, this minimum gradient must be ob-
tained by at least one of the two shocks at each
ventricular site.
Sites of Epicardial Activation After
Subthreshold Shocks
There were 262 unsuccessful shocks, with an average
of 2.31.0 early epicardial sites per shock. The number
of early sites did not differ significantly with the elec-
trode configuration used, as can be seen in Table 2.
Initial activation occurred where gradients were weak
for each electrode configuration (Figures 2F and 3C). In
general, early sites occurred where the gradients were
less than 10 V/cm (Figure 4). There was no evidence of
early epicardial activation occurring in high gradient
areas after unsuccessful shocks. The distribution of
potential gradients at the sites of early activation for all
unsuccessful shocks was approximately normal with a
skew to the left (Figure 5). The mean potential gradient
at sites of early activation after unsuccessful shocks was
9.55.4 V/cm; median was 7.9 V/cm. Ninety-eight per-
cent had gradients less than 25 V/cm, and 87% had
gradients less than or equal to 15 V/cm. Fifty-two
percent of the earliest sites were within 1 V/cm of the
minimum measured potential gradient for the corre-
sponding shock. The largest discrepancy between min-
imum measured gradient and the gradient at the site of
earliest epicardial activation was 21.6 V/cm (7.7 and
29.3 V/cm, respectively), occurring with an unsuccessful
sequential shock. Mean gradient at the sites of early
activation for each electrode configuration after the
highest subthreshold shocks is shown in Table 2.
Sites of Epicardial Activation After Successful Shocks
Of the 67 successful shocks, 11 (16%) were followed
by the immediate resumption of supraventricular
rhythm (type A defibrillation) and 55 (83%) were
followed by one to several ventricular ectopic complexes
before the resumption of supraventricular rhythm (type
B defibrillation).5"18"19 Type A defibrillation occurred in
six of 19 R: V shocks, none of 14 L: V shocks, four of 18
(R+L):V shocks, and one of 16 R:V->L:V shocks. Of
the 55 defibrillations followed by transient ventricular
 Wharton et al Defibrillation Gradients 1517

A. c
R R:V R (R+L):V

D
B. R R:V -> L:V
L :V
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017

v v
FIGURE 4. Maps indicate sites of epicardial activation after all unsuccessful shocks between 4 J and the highest subthreshold
shockforR: V (single anode-single cathode configuration) (panel A), L: V (single anode-single cathode configuration) (panel B),
(R+L): V (dual anode-single cathode configuration) (panel C), and R: V-L: V (sequential 7-msec shocks separated by 1 msec
given to R: Vand L: V) (panel D). Relative locations of shock electrodes are indicated by double cross-hatched oval areas. Site(s)
of earliest activation are indicated with filled circles; other sites of early activation are indicated with filled triangles. Small branch
arteries are for illustrative purposes only and do not correspond to actual location of branch arteries. Cross-hatched and stippled
areas on each figure represent epicardial surface in which gradients less than 10 V/cm were measured during the highest
subthreshold shocks for the majority (greater than six) and the minority (less than or equal to six) of experimental animals,
respectively. Solid line represents border of area in which at least one dog had gradients less than 10 V/cm for the initial 4-J shocks.
Distance between solid line and apical border ofstippled area illustrates decrease in low gradient area needed to reach the highest
subthreshold shock. The much greater difference for L: Vshocks (panel B) than for other configurations between the solid line (4-I
shocks) and apical border of the stippled area (highest subthreshold shocks) is because of the much greater defibrillation threshold
for L: Vshocks. Figures graphically illustrate that early sites of activation for all electrode configurations occur mainly in areas in
which gradients are less than 10 V/cm and demonstrate similarity of generated fields with each electrode configuration within all
experimental animals. Note that despite the fact that L: Vhad the highest defibrillation threshold, epicardial surface area of lowest
gradients is smallest. Also note that for R: V shocks (panel A) but not (R+L): V shocks (panel C) that gradients are frequently
somewhat greater than 10 V/cm on the posterior right ventricle and that early sites of activation do not occur in this area for R: V
shocks but do for (R+L): V shocks. This is less clearly demonstrated for R: V-L: V shocks (panel D) perhaps because the
compositefield is used to generate the figure. Occurrence of several early sites of activation outside of cross-hatched or stippled areas
is caused primarily by their occurrence after shocks of considerably less total delivered potential (and larger area of low gradient)
than the highest subthreshold shock.

ectopic activity, there were 94 early sites or a mean of
1.70.8 per successful shock. Ninety (96%) of these
early sites occurred in areas in which the gradient was
weak; however, four (4%) occurred in high gradient
areas (mean gradient, 7140 V/cm) at the perimeter of
a shock electrode after L:V defibrillations only. Mean
gradient at the site of origin of epicardial activity after
all successful shocks followed by ectopic activity was
15.4+15.4 V/cm (12.9+6.9 V/cm if the four high gradi-
ent sites are omitted).
Eleven episodes of nonsustained, monomorphic ven-
tricular tachycardia occurred after defibrillating shocks.
Eight of these episodes occurred after a pause of
167-852 msec after one to seven postshock, ectopic
activations originating in low gradient areas (Figure 6).
One episode occurred approximately 8 seconds after a
successful type B shock, with 6.5 seconds of intervening
sinus rhythm before the initiation of ventricular tachy-
cardia. The remaining two episodes occurred 81 and 328
msec after type A defibrillating shocks. All episodes of
ventricular tachycardia arose from or near the perime-
ter of ventricular shock electrodes in which gradients
were high (Figure 6). Mean gradient at the site of
earliest epicardial activation of these ventricular tachy-
cardias was 47.3+15.5 V/cm. Evidence supporting mac-
roreentry was not identified for any of these episodes of
ventricular tachycardia.
Myocardial and Electrode Impedance
Electrode impedance accounted for approximately
one third to one fourth of total impedance for all
 1518 Circulation Vol 85, No 4 April 1992

TABLE 2. Potential Gradients and Number of Sites of Initial Activation for Subthreshold Shocks 10% or Less Than
Defibrillation Threshold Energy
R:V L:V (R+L):V R:V->L:V
Minimum gradient (V/cm) 5.81.6 7.31.6 5.71.5 6.61.7*
Number of early sites 1.90.9 1.40.5 2.21.0 2.00.6
Gradient at sites of 9.04.5 12.26.7 11.85.9 12.66.7*
initiation (V/cm)
Maximum gradient (V/cm) 105.966.0 181.897.1t 130.197.5 99.352.4*
R:V and L:V, single anode-single cathode configurations; (R+L):V, dual anode-single cathode configuration;
R V-L: V, sequential 7-msec shocks separated by 1 msec given to R: V and L: V.
*Represents field parameter from maximum combined gradient field (see text).
tp<0.05 with respect to R:V-*L:V.

electrode configurations. The mean electrode, cardiac,
and total impedance at DFT was generally less with
(R+L):V compared with the other configurations (Ta-
ble 3), although the impedance for (R+L):V was
indirectly calculated (see "Methods"). The smaller elec-
trode impedance for (R+L): V was presumably due to
the larger combined anodal surface area. At DFT,
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
from multiple sites, which was not done in the above
studies. Chen et a14 have attempted to directly measure
the distribution of potentials generated across most of
the ventricles by defibrillation electrodes; however, the
fields recorded were from 1-2-V shocks delivered dur-
ing the fully repolarized period of sinus rhythm. Com-
parison of the potential gradient field generated by the

electrode, cardiac, and total impedance were similar
between R: V and the R: V component of the sequential
shock; however, all components of impedance were
significantly higher for the L: V component of the
sequential shock than for L: V shocks alone (p<O.O5).
Because impedance decreased as shock strength was
increased (see "Appendix"), the lower mean imped-
ances for the L: V shocks, compared with the L: V
component of the sequential shock, may simply reflect
the larger shock strength at DFT for the single L: V
shocks. When L V shocks of similar voltage to the L: V
component of the sequential shock were compared,
impedances were not significantly different.
Discussion
Defibrillation presumably depends on achieving a
sufficient transsarcolemmal potential change across all,
or most, of the ventricular myocardium. The transsar-
colemmal potential change should be related to the
low-voltage shocks with the activation sequences re-
corded immediately after high-voltage defibrillation
shocks showed that the continuation of ventricular
fibrillation after unsuccessful shocks arose in areas in
which the field was weak during low-voltage shocks.
Because recordings were not made during high-voltage
shocks, quantitative field requirements for defibrillation
could not be defined. Furthermore, impedance and the
fraction of delivered potential appearing across the
ventricles are markedly different for 1-2-V shocks than
for shocks of hundreds of volts (see "Appendix").
Mapping system modifications now allow the mea-
surement of high-voltage defibrillation shocks as well as
myocardial activation before and after shocks.5"4 Using
such a device, Witkowski et a15 recently showed that
sites of early activation after unsuccessful high-voltage
shocks occurred in areas in which the field is weak. By

.rfl-1
extracellular potential gradient or current density.120 140
There is some indirect evidence to support the theory
that a minimum potential gradient or current density is
required across all or most of the myocardium to 120
achieve defibrillation. The efficacy of different paddle
sizes in transchest defibrillation and of two different
external shock electrode locations was shown to be
) uu
innj
r
proportional to the generated intracardiac potential u- 80)I
0
U)
gradients.2'3 However, the limited number of electrodes W
used in these studies did not allow determination of the :~> 60
myocardial distribution of gradients or of the possibility z
of a minimum gradient necessary for defibrillation.
Geddes et al,2' using a presumed uniform current 40
density field across an excised, perfused, whole heart in
vitro, showed that DFTs in terms of current density 20
were similar for three different waveforms of equal
duration. From these studies, Geddes et al estimated V_ W
0 41
A
8 12
1-
16 20 24
^ .
>28
-

that a minimum potential gradient of approximately 10 GRADIENT (V/cm)

V/cm was necessary for defibrillation with 10-msec
rectangular or trapezoidal waveforms, comparable with
the waveform used in this study.
Because the heterogeneous and anisotropic proper-
ties of the heart preclude uniform changes in potential,
determination of the potential field produced across the
ventricles by the shock requires recording potential
FIGURE 5. Bar graph shows distribution of gradients from all
electrode configurations measured at sites of earliest postshock
activation (shaded area) and from ohr earl sites of epicardial
activation (unshaded area) aftershocks. Eariest and other earl
sites were included from all unsuccessful shocks. Group
mean+SD=9.5+5.4 V/cm; range, 1. 7-35.2 V/cm.

A TABLE 3. Electrode, Cardiac, and Total Impedance at
(R+L): V
RVOT
1 T----t- -r- 1
APEX
500ms.c
B sequential 7-msec shocks separated by 1 msec given to R : V and L: V.
14
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
GRADIENT FIELD
c
..A 4. 474
4~~~~~~~~~6
WRIEST SITE
EPICARDIAL ACTIVATION
FIGURE 6. Panel A: Two-channel recording from right ven-
tricular outflow tract (RVOT) and left ventricular apex adjacent
to cathode is shown for a successful 284-V shock with the
(R+L): V (dual anode-single cathode configuration) shock
from a different experimental animal from the one shown in
Figures 2 and 3. Recordings show approximately 500 msec of
ventricular fibrillation, which is interrupted by the shock. A
single repetitive complex occurs after the shock that had its origin
in the base of the right ventricle near the RVOT electrode. This
is followed by a regular uniform complex tachycardia with a
cycle length of approxunately 250 msec and with its origin in the
left ventricular apex. Although recordings are bipolar after
shock, note prominent ST segment elevation at apical recording
site, which is suggestive of myocardial injury. Panel B: Map
shows potential gradientfield for the same shock. Panel C: Map
of second activation sequence after defibrillation shock shown in
panel A shows origin of ventricular tachycardia from electrode
adjacent to apical shock electrode. Isochronal lines are 10 msec
apart. Calculated gradient at this site was 57 V/cm (panel B).
Evidence of macroreentry in epicardial electrodes was not
identified. Rapid activation of most of the apex of the heart
suggests an origin deep within ventricles such as from endocar-
dium or septum.
determining the potential gradient field at DFT for
several different electrode configurations, the present
study provides direct evidence that a minimum potential
Wharton et al Defibrillation Gradients 1519
Defibrillation Threshold
Electrode Cardiac Total
-~
Anode: Cathode (fQ) (Q1) (fQ)
R:V 34.4+9.3 88.8+23.9 123.328.0
L:V 28.410.1 79.712.2 108.1 18.5
(R+L):V 27.3 7.0 71.915.0 99.2 19.6
R: V-> 32.97.7 89.520.2 122.424.0
L: V 37.5 + 11.4 93.4 16.6 130.8+22.4
R:V and L: V, single anode-single cathode configurations;
(R+L):V, dual anode-single cathode configuration; R:V->L:V,
gradient field is necessary over both ventricles for
defibrillation. Despite the marked differences in the
electrode configurations used and the differences in
their corresponding fields, all obtained similar minimum
gradients of approximately 6-7 V/cm at their DFTs.
This is similar to the minimum required potential
gradient estimated by Geddes et a121 and to that re-
ported by Witkowski et al.5 Despite the marked differ-
ence in total current, voltage, and energy required for
defibrillation by L: V compared with the other configu-
rations, the minimum gradients were not significantly
different. Also, the efficacy of R: V->L: V appears to be
related to its ability to achieve this minimum gradient
over both ventricles during at least one component of
5- the sequential shock, although the minimum gradient of
either component shock may be considerably less. Thus,
it appears that a minimum potential gradient is neces-
sary across all of the ventricular myocardium to achieve
defibrillation. Unfavorable electrode configurations re-
quire large expenditures of energy or current to obtain
this minimum gradient at sites in which their field
strength is weak.
The fields generated by electrodes with small surface
areas relative to that of the heart are inhomogeneous, as
illustrated by the fields generated by all of the electrode
configurations used in this study. Because most of the
potential is lost near the shock electrodes, high poten-
tials (and thus high energy or current) must be deliv-
ered to generate gradients distant to the shock elec-
trodes that are sufficient for defibrillation. Electrodes
with larger surface areas, such as those used clinically,
should improve the distribution of gradients during
shocks and improve defibrillation parameters.2 Small
electrodes were used in this study to generate uneven
gradient fields to enhance discrimination of defibrilla-
tion field requirements and to allow sufficient distribu-
tion of epicardial recording electrodes for precise field
and activation mapping. The finding that a minimum
gradient of 6-7 V/cm is needed for defibrillation should
also hold true for larger electrodes.
The DFT was not improved by coupling two anodes
together with (R+L):V shocks. Chang et at22 reported
similar results with the use of an intracardiac catheter
and either an epicardial or subcutaneous patch elec-
trode. In the present study, the potential gradient field
of the (R+L):V shocks was weak at the base of the
anterior right ventricle, as was true for the fields of the
R: V and L: V shocks. Thus, this particular combination
of electrodes did not improve the distribution of gradi-
ents across the ventricles. It is also possible that during
 1520 Circulation Vol 85, No 4 April 1992
(R+L):V shocks, vector components of current flow
between the R and V electrodes and the L and V
electrodes were oppositely directed, particularly be-
tween the two anodes in the area of the atrioventricular
groove, and that this may have resulted in vector
cancellation of these components.
Sequential shocks with short shock separation times
to two electrode configurations using an intravascular
catheter and patch-electrode configuration or a three-
patch-electrode configuration have been shown to have
greater efficacy than shocks to single pairs of elec-
trodes.9'22-24 However, not all sequential shock elec-
trode configurations significantly alter the energy re-
quirements for defibrillation.22,25'26 The present study
confirmed the greater efficacy of sequential shocks to
two different epicardial electrode configurations. The
mechanism by which sequential shocks may lower the
DFI' is presumably due to improvement in the distribu-
tion of potential gradients across the ventricles. The
L: V component of the sequential shocks at the DFT
significantly decreased (by 52%) the number of record-
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
ing electrodes that measured a gradient less than or
equal to 10 V/cm during the preceding R: V component
of the sequential shock. Thus, the overall surface area of
the ventricles that was exposed to a weak gradient field
during the sequential shocks was substantially de-
creased compared with the R: V shock alone. There was
no evidence that the initial R: V shock changed imped-
ance to the subsequent L: V shock to enhance current
delivery of the latter. However, the minimum gradients
of the composite gradient field of the sequential shocks
were greater than the minimum gradients of each
component shock but were similar to the minimum
gradients produced by all of the other single-shock
electrode configurations at the DFT. This again suggests
the necessity of obtaining a certain minimum gradient
over all of the heart for defibrillation. Greater increases
in defibrillation efficacy may be obtained by designing
electrode configurations in which one shock field is
strong where the other is weak, and neither is weak in
the same area.
None of the electrode configurations examined in this
study produced adequate fields along the right ventric-
ular base, especially along the right ventricular outflow
tract. One explanation for this finding is that the three
electrodes describe a plane that is located posteriorly
(Figure 1B), so that the right ventricular outflow tract,
which bulges anteriorly, lies the greatest distance from
this plane to produce an area in which the field is weak.
This effect may not be as pronounced in the human
heart because the right ventricular outflow tract does
not project as far anteriorly relative to this plane.
The present study confirms and expands on the
finding that ventricular fibrillation first appears after
unsuccessful shocks in areas in which the field is weak,
regardless of the electrode configuration used.4'5
Eighty-seven percent of gradients at the sites of early
epicardial activation after unsuccessful shocks were less
than 15 V/cm, and 52% were within 1 V/cm of the
minimum measured gradient. Sites of initiation of ven-
tricular fibrillation in high gradient areas after unsuc-
cessful shocks were not seen. The lack of a closer
correlation between the absolute minimum potential
gradient generated by a field and the gradient at the site
of resumption of ventricular fibrillation may be ex-
plained by a number of factors, such as measurement
error and the lack of intramyocardial and endocardial
recording sites.
Another factor explaining why earliest activation
after an unsuccessful shock is not always in the region
exposed to the weakest potential gradient is variation in
the electrophysiological state of the myocardium at the
time of the shock. Besides the spatial distribution of
potential gradients and the temporal waveform of the
shock, major determinants of the effect of a shock are
probably the state of myocardial activation, refractori-
ness, and wavefront propagation at the time of the
shock. Such factors may prevent the postshock appear-
ance of ventricular fibrillation activation fronts at many
sites whether or not the potential gradient of the shock
exceeds a minimum threshold. For example, when the
region in which the shock field is weakest is in its
absolute refractory period at the time of the shock,
fibrillation will not reappear from this region even when
the potential gradient is below threshold. At another
point in time during fibrillation, this region may not be
absolutely refractory, and fibrillation activation fronts
may originate from this region after the shock,19'27
assuming that adjacent sites are capable of conducting
the impulse and that surrounding wavefronts do not
collide and terminate the reentrant circuit. Consider-
ation of these interactions may help to explain the
observation that there exists a range of voltages or
energies near threshold in which defibrillation is best
described in terms of a probability function.28'29 The
occurrence of a shock strength with which defibrillation
is always achieved and an upper limit of shock strength
that will not induce fibrillation no matter when it is
introduced into the vulnerable period27'30'31 suggests
that there should exist a potential gradient above which
fibrillation will be terminated regardless of the state of
the myocardium. However, because the shock strength
that is always successful is usually greater than that at
the DFT,28,29,31 the minimum gradient measured in this
study is probably less than that required to always
defibrillate.
The occurrence of a minimum gradient at the DFT is
most consistent with either the total ventricular depo-
larization (or extinction) hypothesis of defibrillation32 or
the upper limit of vulnerability hypothesis,19'27'30 be-
cause both theories predict a minimum required poten-
tial gradient across all of the ventricular myocardium.
The finding of a required minimum gradient across all
of the ventricular myocardium is less consistent with the
critical mass hypothesis.18'33 The latter hypothesis pre-
dicts that the minimum necessary gradient for different
electrode configurations should be equal, not over the
entire ventricular myocardium but at the boundary
enclosing the critical mass. The absolute minimum
gradient would not necessarily have to be equal for
different electrode configurations, because it would
depend only on the rate of decline in the gradient field
beyond the border of the critical mass. The critical mass
for defibrillation has been estimated to be approxi-
mately 75% of the ventricular myocardium with pro-
longed depolarization with potassium administration.33
If so, equal potential gradients should be found at the
border of the regions enclosing 75% of the ventricular
myocardium for different electrode configurations, not
at the minimum gradient value over all of the ventricles.

The findings in this study that 1) the L: V shock had a
smaller surface area of low gradients compared with the
other shocks (Figure 4), despite the fact that it was
markedly less efficient in defibrillation, and 2) the
minimum potential gradients are not significantly differ-
ent for the different lead configurations both run con-
trary to the predictions of the critical mass hypothesis.
However, because endocardial and septal gradient mea-
sures are not available for determination of the three-
dimensional potential gradient field, it is possible that
intramyocardial potential gradient distributions do not
parallel in a relatively symmetrical manner the distribu-
tion measured on the epicardium and that a critical
mass less than 100% existed but was not identified in
this study. We hope that further developments in the
calculation of gradients across the entire volume of the
heart will help to resolve this issue.34
A potential gradient greater than approximately 5
V/cm for a 3-msec, low-tilt, monophasic shock delivered
during the relative refractory period of a passing wave-
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
front induces conduction block, presumably by prolong-
ing refractoriness; gradients less than this do not pro-
duce conduction block.35 This critical interaction
between refractoriness and decreasing field strength
may lead to the induction of reentry and ventricular
fibrillation. According to the upper limit of vulnerability
hypothesis, subthreshold shocks will halt all fibrillation
wavefronts but reinitiate ventricular fibrillation, pre-
sumably by a mechanism similar to that just described.19
The similarity between the minimum gradient of 6-7
V/cm for defibrillation found in this study and the
critical gradient of 5 V/cm for electrical induction of
fibrillation in the vulnerable period of regular rhythm is
consistent with but does not prove this hypothesis.
There are several limitations to this study. Wave-
forms of depolarization are obscured during defibrilla-
tion by the shock and immediately after the shock by the
baseline deflection generated by switching off the map-
ping system modification; thus, depolarization during or
immediately after the shock cannot be identified or
mapped. Furthermore, the lack of septal, intramyocar-
dial, and endocardial recording sites limits the resolu-
tion of the mapping and the determination of all
components of the gradients. Specifically, the lack of
transmural recordings may decrease the absolute value
of the measured gradients. For low-voltage shocks to an
electrode array resembling R: V, transmural gradients
in the right ventricular outflow tract are approximately
half the tangential gradient.4 The exposure of the
anterior surface of the heart to air in the open-chest dog
preparation may increase the anterior gradient mea-
surements due to a boundary effect; however, the
greater tangential current flow would improve gradient
calculation using epicardial recording electrodes. Mea-
surement of rapidly changing gradients near shock
electrodes requires small interelectrode distances to
maintain accuracy; irregularities in high gradient areas
in this study may reflect inadequate spacing of recording
electrodes or inhomogeneous current distribution
around the perimeter of the shock electrode. Depolar-
ization of ventricular myocardium is dependent on fiber
orientation relative to the vector of the potential gradi-
ent.36 Whether fiber orientation affects the field require-
ments for defibrillation or whether different areas of the
Wharton et al Defibrillation Gradients 1521
ventricles have the same field requirements were not
assessed in this study.
When only a certain minimum gradient is needed for
defibrillation, electrode configurations that must gener-
ate very high gradients near the shock electrodes to
achieve this minimum gradient distantly not only waste
energy but increase the risk of inducing variable degrees
of myocardial injury. High potential gradients may
cause cell injury and death, decreased contractility,
asystole, conduction block, and dysrhythmias.'37-42 Ex-
tremely high gradients may even generate intractable
ventricular fibrillation secondary to induced injury.30,40
Despite the great unevenness of the fields generated by
the electrode configurations used in this study, post-
shock arrhythmias occurred relatively infrequently.
There was no evidence that the high gradients at the
shock electrodes directly initiated ventricular fibrillation
after unsuccessful shocks, although repetitive activity
occurred infrequently in high gradient areas after suc-
cessful shocks. Specifically, this occurred only after
successful L: V shocks, which generated the highest
gradients for defibrillation, perhaps reaching a specific
injury threshold. High-voltage shock induction of ven-
tricular fibrillation (type II ventricular fibrillation)
probably requires voltages higher than those used in this
study.3040 Although shock-induced arrhythmias may oc-
cur clinically41 and have been shown to occur more
frequently after high-voltage shocks in vitro,'37'40 the
present study is the first to map the origin of postshock
ventricular tachycardia from the site of the shock elec-
trode, which in all cases was recorded from electrodes at
the perimeter of the shock electrode possibly reflecting
the higher current density at the electrode edge.43 The
mechanism of the postshock ventricular tachycardia
seen in this study could not be ascertained; evidence of
epicardial macroreentry was not seen.
The ability to measure shock potentials also allows
measurement of various components of impedance,
such as that at the electrode-myocardial interface or
across the heart itself (see "Appendix"). The imped-
ance measured in this study is generally greater than
that reported by others for transmyocardial shocks,44,45
presumably reflecting the small size of the shock elec-
trodes used in this study. Unlike a previous study,27 the
present study did not show significant changes in im-
pedance with increasing shock voltage for R: V,
(R+L): V, or the components of R:V-L: V. This may
be due to the random sequence of shock delivery and/or
to the relatively small range of shock voltages delivered
to these configurations. There was a significant change
with L: V, which, because of its decreased efficacy,
required a larger number of shocks over a greater range
of voltages to achieve a DFT. When R: V shocks were
given over a greater range of voltages, marked changes
in impedance were noted (see "Appendix"). However,
Lawrence et al44 were not able to show a change in
impedance between shocks of different strengths when
more than 5 minutes elapsed between shocks and
showed only a very small change when the shocks were
separated by a few seconds.
The ability to map the potential gradient field gener-
ated by shock electrodes should lead to marked im-
provements in their design and function. If the desired
goal is a uniform field slightly exceeding a certain
minimum gradient, the possible electrode combinations
 1522 Circulation Vol 85, No 4 April 1992
are almost limitless; however, application of the rule of
superposition may lead to a more efficient way of testing
potential fields for numerous electrode configurations
with a relatively limited number of shocks. Further
description of factors involved in electrode impedance
may also result in marked improvement in electrode
design. If electrodes, electrode configurations, and
shock sequences can be designed that will generate a
more even field that only slightly exceeds the minimum
gradient required for reliable defibrillation without in-
creasing electrode impedance, implantable defibrilla-
tors then can be designed that will require less energy to
operate. This in turn will have major benefits, such as
increasing the lifetime and decreasing the size of the
battery of implanted units and decreasing the pain,
potential cardiac dysfunction, and dysrhythmias associ-
ated with defibrillation shocks.
Appendix
Effect of Shock Strength on Impedance
In an earlier study4 with low-voltage shocks from 1 to 2 V
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
given to R:V using 4.5-cm2 titanium mesh electrodes, only
173% of the delivered shock potential was measured on the
ventricles, whereas in the present study, recorded myocardial
potential was 76.46.3% of that delivered with 2.64-cm2
electrodes. The addition of recording electrodes beneath the
shock electrodes and inclusion of the voltage drop across the
atria in this study may explain some but not all of this
difference. To see if part of this discrepancy was caused by
differences in impedance for low- and high-voltage shocks, an
additional dog was studied with the same surgical preparation
as described in "Methods." However, 4.5-cm2-round titanium
mesh shock electrodes with a recording button electrode
underneath were attached in an R: V configuration. Shocks of
14-msec duration were delivered once each at increasing
strengths as follows: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12.5, 15, 20, 25,
50, 75, 100, 150, 200, 300, 400, 500, 750, and 1,000 V. Shocks
from 1 to 25 V were delivered by a constant voltage device and
from 25 to 1,000 V by the device used in the present study. Two
calibrated input dynamic ranges (10 V and 500 V) to the
amplifiers were used to record potentials. Current was mea-
sured on an oscilloscope.
With 1-4-V shocks, total impedance increased markedly from
254 to 434Q1, decreased abruptly to 273Q1 at 5 V, and then
gradually declined with increasing potential down to 106 fl at
1,000 V. A plot of the logarithm of potential delivered versus the
calculated total impedance illustrates the marked change in
impedance at low voltages (Figure 7). Beyond approximately 10
V, the relation between impedance and the logarithm of deliv-
ered voltage became more linear.
To define where the change in impedance occurred, mean
differences in potential between successive rows of electrodes
starting at the apical shock electrode were calculated and
divided by the total measured current as an estimate of
effective impedance of the layers of ventricular myocardium
between each row of electrodes. The greatest impedance was
calculated between the shock electrode and the recording
electrode underneath it (row 1) for shocks of 1-20 V (Figure
7). At 25 V, however, the impedance at row 1 and the next
most cephalad row (row 2) equalized with further increases in
delivered potential causing the greatest impedance change at
row 2, principally because impedance at row 1 continued to
decrease, whereas it was relatively stable at row 2. The marked
increase in impedance at 4 V is reflected in the impedance
curves at all levels of the ventricles, although most of this
occurs in the apical third of the myocardium in rows 1-3
(Figure 7).
For a 1-V R: V shock, 36% and 15% of delivered potential
was measured between electrodes beneath (row 1) and imme-
400
: 3000
U
z 4V
< 200
Lin
100
F
1
2 -25V
3 -
o 4-7
1 V10 100 100J
V
1VOL 100(V 1000 v
VOLTAGE (V)
FIGuRE 7. Graph of calculated total impedance or mean
change in potential divided by total delivered current between
successive rows of electrodes across the ventricles as a function
of logarithm of delivered potential from R: V (single anode-
single cathode configuration) shocks of 1-1,000 Logarith- V
mic scale wasused to emphasize abrupt changes at low
delivered potential. Line labeled "Total" represents relation of
total impedance and delivered potential. Lines labeled 1-7
represent changes between successive rows of electrodes pro-
gressively cephalad to apical shock electrode. Thus, line 1
represents impedance between shock electrode and recording
electrode immediately underneath (row 1). Line 2 represents
estimated impedance changes between recording electrode un-
derneath apical shock electrode and ring of electrodes at
perimeter of apical shock electrode (row 2). Lines 3-7 represent
estimated impedance changes between successive rows progres-
sively cephalad to recording electrodes at perimeter of shock
electrode. For further discussion, see 'Appendix."
diately encircling (row 2) the apical shock electrode, respec-
tively, and the electrode beneath the right atrial anode. These
measurements were not significantly changed at 4 V and were
in accordance with the previous study.4 As delivered potential
increased from 5 V, 25 V, 100 V, and 1,000 V, the percent of
delivered potential measured in row 1 steadily increased to
48%, 66%, 73%, and 75% and in row 2 from 20%, 32%, 34%,
and 43%, respectively.
The current waveform for shocks less than V showed a 5
logarithmic increase in current flow during the period of the
shock, suggesting a large contribution of capacitance to the
overall impedance. This capacitive element may explain the
initial increase in total impedance. As potential was increased
above 5 V, the current waveform showed progressively less
distortion and impedance decreased, primarily at or near the
electrode-myocardial interface. This strong capacitive element
to impedance may be generated to some extent by electrode
polarization at the myocardial interface during low-voltage
shocks. However, the occurrence of similar changes in esti-
mated impedance distant to the apical shock electrode implies
myocardial changes occurring as a function of increasing
potential delivered. This capacitive component of impedance
progressively decreased with increasing delivered potential so
that in the range of potentials necessary for defibrillation with
epicardial electrodes, impedance appears to be primarily
resistive, confirming the work of others.45
Acknowledgments
We would like to thank Joseph C. Greenfield Jr. for his
support and advice; and Ellen G. Dixon, Sharon B. Melnick,

Yohannes Afework, Alton T. Ledford, and Cloyce M. Lassiter
for their technical assistance.
References
1. Lepeschkin E, Jones JL, Rush S, Jones RE: Local potential
gradients as a unifying measure for thresholds of stimulation,
standstill, tachyarrhythmia and fibrillation appearing after strong
capacitor discharges. Adv Cardiol 1978;21:268-278
2. Hoyt R, Grayzel J, Kerber RE: Determinations of intracardiac
current in defibrillation: Experimental studies in dogs. Circulation
1981;64:818-823
3. Kerber RE, Hoyt R, Aronson A, Kieso R, Melton J: Evaluation of
a new defibrillation pathway: The tongue-epigastric route:
I. Experimental studies in dogs. JAm Coll Cardiol 1983;2:966-972
4. Chen P-S, Wolf PD, Claydon FJ, Dixon EG, Vidaillet HJ Jr,
Danieley ND, Pilkington TC, Ideker RE: The potential gradient
field created by epicardial defibrillation electrodes in dogs. Circu-
lation 1986;74:626-636
5. Witkowski FX, Penkoske PA, Plonsey R: Mechanism of cardiac
defibrillation in open-chest dogs with unipolar DC-coupled simul-
taneous activation and shock potential recordings. Circulation
1990;82:244-260
6. Babbs CF: Effect of pentobarbital anesthesia on ventricular
defibrillation threshold in dogs. Am Heart J 1978;95:331-337
7. Galindo AH, Davis TB: Succinylcholine and cardiac excitability.
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017
Anesthesiology 1962;23:32-40
8. Worley SJ, Ideker RE, Mastrotatoro J, Smith WM, Vidaillet HJ Jr,
Chen P-S, Lowe JE: A new sock electrode for recording epicardial
activation from the human heart: One size fits all. PACE 1987;10:
21-31
9. Jones DL, Klein GJ, Kallok MJ: Improved internal defibrillation
with twin pulse sequential energy delivery to different lead orien-
tations in pigs. Am J Cardiol 1985;55:821-825
10. Bourland JD, Tacker WA Jr, Geddes LA: Strength-duration
curves for trapezoidal waveforms of various tilts for transchest
defibrillation in animals. Med Instrum 1978;12:38-41
11. Babbs CF, Whistler SJ, Yim GKW: Temporal stability and preci-
sion of ventricular defibrillation threshold data. Am J Physiol
1978;235:H553-H558
12. Hoffman BF, Suckling EE, Brooks CM: Vulnerability of the dog
ventricle and effects of defibrillation. Circ Res 1955;3:147-151
13. Barr RC, Spach MS: Sampling rates required for digital recording
of intracellular and extracellular cardiac potentials. Circulation
1977;55:40-48
14. Wolf PD, Wharton JM, Wilkinson CD, Smith WM, Ideker RE: A
method of measuring cardiac defibrillation potentials, in Proceed-
ings of the 39th Annual Conference on Engineering in Medicine and
Biology, 1986, p 4
15. Wolf PD, Danieley ND, Ideker RE, Smith WM: A digital tape
recorder for electrophysiologic waveforms, in Proceedings of the
38th Annual Conference on Engineering in Medicine and Biology,
1985, p 124
16. Smith WM, Ideker RE: Computer techniques for epicardial and
endocardial mapping. Prog Cardiovasc Dis 1983;26:15-32
17. SAS/STAT User's Guide, Release 6.03 Edition. Cary, NC, SAS Insti-
tute, Inc, 1988
18. Mower MM, Mirowski M, Spear JF, Moore EN: Patterns of
ventricular activity during catheter defibrillation. Circulation 1974;
49:858-861
19. Chen P-S, Shibata N, Dixon EG, Wolf PD, Danieley ND, Sweeney
MB, Smith WM, Ideker RE: Activation during ventricular defibril-
lation in open-chest dogs: Evidence of complete cessation and
regeneration of ventricular fibrillation after unsuccessful shocks.
J Clin Invest 1986;77:810-823
20. Krassowska W, PiLkington TC, Ideker RE: Periodic conductivity as
a mechanism for cardiac stimulation and defibrillation. IEEE Trans
Biomed Eng 1987;34:555-560
21. Geddes LA, Niebauer MJ, Babbs CF, Bourland JD: Fundamental
criteria underlying the efficacy and safety of defibrillation current
waveforms. Med Biol Eng Comput 1985;23:122-130
22. Chang MS, Inoue H, Kallok MJ, Zipes DP: Double and triple
sequential shocks reduce ventricular defibrillation threshold in
dogs with and without myocardial infarction. J Am Coll Cardiol
1986;8:1393-1405
Wharton et al Defibrillation Gradients 1523
23. Jones DL, Klein GJ, Guiraudon GM, Sharma AD, Kallok MJ,
Bourland JD, Tacker WA Jr: Internal cardiac defibrillation in man:
Pronounced improvement with sequential pulse delivery to two
different lead orientations. Circulation 1986;73:484-491
24. Bourland JD, Tacker WA Jr, Wessale JL, Kallok MJ, Graf JE,
Geddes ME: Sequential pulse defibrillation for implantable
defibrillators. Med Instrum 1986;20:138-142
25. Wetherbee JN, Chapman PD, Bach SM Jr, Troup PJ: Sequential
shocks are comparable to single shocks employing two current
pathways for internal defibrillation dogs. PACE 1988;11:696-703
26. Mehra R, Maracaccini S: Comparison of sequential and simulta-
neous pulse defibrillation threshold with a non-epicardial elec-
trode system (abstract). Circulation 1986;74(suppl II):II-184
27. Chen P-S, Shibata N, Dixon EG, Martin RO, Ideker RE: Com-
parison of the defibrillation threshold and the upper limit of
ventricular vulnerability. Circulation 1986;73:1022-1028
28. Gold JH, Schuder JC, Stoeckle H: Contour graph for relating
percent success in achieving ventricular defibrillation to duration,
current, and energy content of shock. Am Heart J 1979;98:207-212
29. Davy JM, Fain ES, Dorian P, Winkle RA: The relationship
between successful defibrillation and delivered energy in open-
chest dogs: Reappraisal of the "defibrillation threshold" concept.
Am Heart J 1987;113:77-84
30. Lesigne C, Levy B, Saumont R, Birkui P, Bardou A, Rubin B: An
energy-time analysis of ventricular fibrillation and defibrillation
thresholds with internal electrodes. Med Biol Eng 1976;14:617-622
31. Wharton JM, Richard VJ, Murry CE Jr, Dixon EG, Reimer KA,
Meador J, Smith WM, Ideker RE: Electrophysiologic effects in
vivo of monophasic and biphasic stimuli in normal and infarcted
dogs. PACE 1990;13:1158-1172
32. Wiggers CJ: The physiologic basis for cardiac resuscitation from
ventricular fibrillation: Method for serial defibrillation. Am Heart J
1940;20:413-422
33. Zipes DP, Fischer J, King RM, Nicoll A, Jolly WW: Termination
of ventricular fibrillation in dogs by depolarizing a critical amount
of myocardium. Am J Cardiol 1975;36:37-44
34. Tang ASL, Wolf PD, Claydon FJ III, Smith WM, Pilkington TC,
Ideker RE: Measurement of defibrillation shock potential distri-
butions and activation sequences of the heart in three dimensions.
Proc IEEE 1988;76:1176-1186
35. Frazier DW, Wolf PD, Wharton JM, Tang ASL, Smith WM,
Ideker RE: Stimulus-induced critical point: Mechanism for the
electrical initiation of reentry in normal canine myocardium. J Clin
Invest 1989;83:1039-1052
36. Frazier DW, Krassowska W, Chen P-S, Wolf PD, Dixon EG, Smith
WM, Ideker RE: Extracellular field required for excitation in
three-dimensional anisotropic canine myocardium. Circ Res 1988;
63:147-164
37. Jones JL, Lepeschkin E, Jones RE, Rush S: Response of cultured
myocardial cells to countershock-type electric field stimulation.
Am J Physiol 1978;235:H214-H222
38. Jones JL, Proskauer CC, Paull WK, Lepeschkin E, Jones RE:
Ultrastructural injury to chick myocardial cells in vitro following
"electric countershock." Circ Res 1980;46:387-394
39. Koning G, Veefkind AH, Schneider H: Cardiac damage caused by
direct application of defibrillation shock to isolated Langendorff-
perfused rabbit heart. Am Heart J 1980;100:473-482
40. Peleska B: Cardiac arrhythmias following condenser discharges
and their dependence upon strength of current and phase of
cardiac cycle. Circ Res 1963;13:21-32
41. Donoso E, Cohn LH, Friedberg CK: Ventricular arrhythmias after
precordial electric shock. Am Heart J 1967;73:595-601
42. Yabe S, Smith WM, Daubert JP, Wolf PD, Rollins DL, Ideker RE:
Conduction disturbances caused by high current density electric
fields. Circ Res 1990;66:1190-1203
43. Kim Y, Fahy B, Tupper BJ: Optimal electrode designs for elec-
trosurgery, defibrillation, and external cardiac pacing. IEEE Trans
Biomed Eng 1986;33:845-853
44. Lawrence JH, Brin KP, Halperin HR, Platia EV, Tsitlik JE, Levine
JH, Guarnieri T: The characterization of human transmyocardial
impedance during implantation of the automatic internal cardio-
verter defibrillator. PACE 1986;9:745-755
45. Kugelberg J: The interelectrode resistances at defibrillation. Scand
J Thorac Cardiovasc Surg 1972;6:274-277
 Cardiac potential and potential gradient fields generated by single, combined, and
sequential shocks during ventricular defibrillation.
J M Wharton, P D Wolf, W M Smith, P S Chen, D W Frazier, S Yabe, N Danieley and R E
Ideker
Downloaded from http://circ.ahajournals.org/ by guest on January 6, 2017

Circulation. 1992;85:1510-1523
doi: 10.1161/01.CIR.85.4.1510
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 1992 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on
the World Wide Web at:
http://circ.ahajournals.org/content/85/4/1510

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally
published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Editorial Office. Once the online version of the published article for which permission is being requested is
located, click Request Permissions in the middle column of the Web page under Services. Further
information about this process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Circulation is online at:

http://circ.ahajournals.org//subscriptions/