Measurement cardiography

of cardiac output by impedance under various conditions

P. BOER, J. C. ROOS, G. G. GEYSKES, AND E. J. DORHOUT MEES Department of Nephrobgy and Hypertension, University Hospital, Utrecht,

BOER, P.,J. C. MEES. Measurement

Roos, GGGEYSKES, AND E. J. DORHOWT of cardiac output by impedance cardiograplzy under wzriuus conditions. Am. J. Physiol. 237(4): H491H496, 1979 or Am, J. Physiol.: Heart Circ. Yhysiol. 6(4): H49lH496, 1979.-The suitability of impedance cardiography as a method to follow changes in cardiac output (CO) was investigated by comparing it with thermodilution and by measurements in conditions with predictable effects on CO. The correlation between absolute CO values obtained by impedance and thermodilution techniques was moderate (r = 0.61), but the percentage changes showed a good correlation (r = 0.84). Headup tilting decreased CO by 25% in 5 normal subjects. Bloodletting in three subjects caused a 20% decrease, reinfusion in one subject a 16% increase. Occlusion of artificial large flow arteriovenous shunts in 6 dialysis patients caused a 12% decrease in CO, whereas the effect was less with moderate and absent with small flow shun&. Ultrafiltration caused a 28% decrease in two dialysis patients. In 20 hypertensive patients CO dropped 25% after salt depletion; propranolol administered intravenously gave a 15% decrease (n = 13), diamxide intravenously a 30% increase (n = 6). Tt is concluded that, provided conditions are strictly defined, impedance cardiography is a reproducible method and suitable for the measurement of intrapatient changes in cardiac output. thoracic impedance plethysmography; noninvasive put measurement; stroke volume; thermodilution put measurement

ONE

OF THE

PARAMETERS

of interest

cardiac outcardiac out-

in clinical

pharma-

cological studies of antihypertensive drugs is the cardiac output (CO). Conventional methods of CO determination, which require invasive techniques, are less suitable for repeated measurements. Being a noninvasive technique, impedance cardiography, as introduced by Kubicek (6, 7) seems to offer an attractive alternative. Although the method is still empirical and the theoretical basis is not very strong, the results show good agreement with those of other, more widely accepted methods (l-4, 8,9). However, a recent publication has again challenged its reliability for clinical use (5). To establish whether the impedance technique is suitable for the detection of changes in CO under divergent conditions, the reproducibility of the method was investigated and the results were compared with the thermodilution technique. In addition, CO was measured by the impedance technique in conditions where a predictable change in CO could be expected: head-up tilting, bloodletting, occlusion of the artificial arteriovenous shunts of dialysis patients, ultrafrlltration of dialysis patients, dehydration by diuretic O~63-6~~5/79/oooO-Oooo$Ol.25

Copyright

0 1979 the American

Physiological

The Nether-la&~

and intravenous injection of propranolol and

therapy,

diazoxide. METHODS

Recordings. An IFM/Minnesota impedance cardiograph, model 304A, was used to estimate stroke volume. A high-frequency (100 kHz) sinusoidal alternating current was applied between the outer pair of four aluminum electrode tapes (3M) placed circumferentially around the subject’s neck, thorax (just below the xiphosternal junction) and abdomen. The fjirst derivative of the changes in impedance with respect to time (picked up from the inner pair of electrodes), together with a phonocardiogram and an electrocardiogram (both for timing purposes) were recorded with a three-channel Siemens Elema Mingograf 34, Blood pressure was measured with a Roche Arteriosonde 1217 automatic blood pressure device. Measurements were made after at least a half-hour rest in the supine position. The subjects were then asked to inspire, the Arteriosonde was switched on, the subjects were asked to exhale, and blood pressure and impedance recordings were made at end-expiratory apnea during six cardiac cycles. For calculations, the mean of five such measurements, taken at 2-min intervals, was used. Thermodilution CO was measured as described elsewhere (10). At least three determinations were made simultaneously with the impedance CO measurements. Cakulations. Stroke volume was calculated from the formula empirically derived by Kubicek et al. (6, 7) SV = p (L/G)’ T (dZ/dt)min in which SV = ventricular stroke volume (ml); p = blood resistivity at 100 kHz (ohm.cm) derived from the hematocrit with a nomogram (Minnesota); L = mean distance between inner pair of electrodes, taken at the anterior and posterior midlines (cm); 20 = basal thoracic impedance between the inner electrodes (ohm); T = ventricular ejection time (s), measured between the zero crossing preceding the minimum peak of dZ/dt at the first highfrequency component of the second heart sound; and (d Z/dt) min = minimum peak value of the impedance derivative (ohm s-’ ) . The other equations used were l

CO = HR x SV/l,oOo in which HR = heart rate (min-‘); output (1 min-‘)

and CO = cardiac

l

TPR = 80 x MAP/CO Society

H491

Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (137.154.019.149) on January 10, 2019.

H492

in which TPR = total peripheral resistance (dyn+ cm-“); and MAP = mean arterial pressure (mmHg) = % systolic + %jdiastolic blood pressure. The cardiac index (CI) and stroke index (SI) were calculated by dividing CO and SV by body surface. For calculations of mean data of groups, CI and SI rather than CO and SV were used. Statistical analyses were performed by Student’s t test or Wilcoxon’s ranking test for paired data and by analysis of variance; correlations were calculated by the method of the least squares or Spearman’s rank correlation test. The following investigations were carried out in various groups of normal subjects and patients, none of them suffering from severe pulmonary or cardiac diseases (like shunts or valvular diseases), with the exception of signs of left ventricular hypertrophy on the ECG in some of the hypertensive patients. Stu’dy I. In four normal male subjects (two of them on two separate occasions), the effects of varying the total distance between the inner electrodes on calculated stroke volume was investigated. In three other normal subjects the effect of varying the electrode distance along the posterior midline only was studied, the midsternal distance being kept constant. Study 11. The reproducibility was assessedby comparing SV measured as described under METHODS at two times in the morning (t = 0 and t = 30 min) in 12 healthy young subjects (6 male, 6 female), who had been supine for at least 1 h. The study was repeated two days later under the same conditions at the same times to keep variation in SV and CO as low as possible. The withinday reproducibility was assessedby comparing SV measured at day I, t = 0 to that at t = 30 and SV at day 2, t = 0 to that at t = 30; the day-to-day reproducibility was assessedby comparing SV measured at day 1, t = 0 to that at day 2, t = 0, and SV at day 1, t = 30 to that at day 2, t = 30. Study 111. The well-known effect of tilting on CO was evaluated on the basis of head-up tilting with a small elevation (20’) in five healthy young male subjects. Study IV. Two patients (I male, 1 female) with persistent hypertension in whom blood volumes were found to be expanded, were subjected to bloodletting of 670 and 460 ml. In addition, in one healthy volunteer, blood was withdrawn (500 ml) and reinfused 2 days later. Recordings were made before, during, and after the changes in blood volume. Study V. In 1IJpatients on chronic hemodialysis treatment (12 male, 6 female) the effect of rapidly alternating opening and closing of the artificial arteriovenous shunt was studied. The shunts were of the following types: Scribner (external, small flow), radialis fistula or Cimino (internal, moderate flow), and brachialis fistula (internal, large flow). Study VZ. In two female dialysis patients the effect of volume depletion by ultrafiltration was studied. Study VZL Twenty patients (18 male, 2 female) with essential hypertension were put on a diet containing 60 meq Na/day and then treated with salt depletion, i.e., 4 days on a diet containing 10 meq Na/day, initiated by 80 mg furosemide three times daily on the 1st day. CO was measured before and on the 4th day of volume depletion

BOER,

KOOS,

GEYSKES,

AND

UORHOUT

ME=ES

in three of the patients, both with impedance cardiogra-phy and thermodiIution. Study VIII. The effect of various therapeutic maneuvers known to influence CO, was studied simultaneously by both impedance and thermodilution techniques. In seven patients with essential hypertension the effect of intravenous injection of 5 mg propranolol was studied (on 13 occasions) and in four patients the effect of intravenous injection of 300 mg diazoxide (on 6 occasions). RESULTS

Study 1. The effect of varying the mean electrode distance on peak height (dZ/dt), thoracic impedance (&), and calculated SV is shown in Fig. IA; the effect of varying the dorsal distance only is shown in Fig. 1B. For both cases an increase in the distance tended to give a rise in dZ/dt and then to level off. As could be expected, Z0 increased linearly with increasing distance. The net effect is an increase in SV with increasing L. Study II. The within-day reproducibility data are shown in Fig. 2A. The SV measured at t = 0 correlated welI with the SV at t = 30 (method of the least squares, r = 0.96, P < 0.001); the differences between t = 0 and t = 30 were not significant (paired t test, P > 0.10, mean difference -LO%, coefficient of variation 4.1%). The dayto-day reproducibility data are shown in Fig. 2B. The SV A

sv h

30-20 ml 2%1 209i .' 15-

1 I #

-' 15

25

35

l

-. 15

25

/,

35

- 15

35

LI/-l-I-l 15

B

2,5-1

dZ/dt (IL.set-'1

1 2p1 IS*IL //-I -1 15

,/.’

l -•

,-I

25

35

L/f-J-1-l 15

25

25

35

back (cd FIG. 1. Effect of varying the distance at the posterior Table 1.

total mean electrode distance L (A) and midline only (B). For abbrevations, see

Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (137.154.019.149) on January 10, 2019.

ME=ASUKEMENT

OF

CARDIAC

OUTPUT

I IOO-

BY

IMPEDANCE

/ ‘3 4

l

0

y = 0.97x + 2.3 r = 0.96 p< 0.001 n=24

l 8 8

’

759

a

%

/

509

tJ/

A ’ L //-I-l50 17%S.V.

day2(

S.V. t=o

-1-l-l Xi0

75 ml)

125

(ml) 150

175

8

I

‘4 o=/ l

/ 4p

0 6

0 1r

/ 0% ma

= I,07 x - 0.5 r = Q97 p{ 0.001 n=24

I

“L //-r-i-

50

-I-

7s

l

y

/ 0-o 50-1 / 0 lA0

S.V-dayl

125

(ml) -I

A0

H493

CARDIOGKAPHY

175

2. Within-day (A) and day-to-day (B) reproducibility of stroke volume measured by impedance cardiography. A: a, day I, t = 0 vs. t = 3O;CJ duy2, t = 0 vs. t = 30. B: l , t = 0, duy 1 vs. day 2; i-1, t = 30, day I vs.duyZ. FIG.

measured at day 1 correlated well with the SV measured at duy 2 (r = 0.97, P < 0.001); the differences between day I and day 2 were not significant (paired t test, P > 0.10, mean difference +1X%, coefficient of variation 4.1%). In addition, as every point in Figs. 2A and 2B was calculated from five series of dZ/dt complexes (as described under METHODS); analysis of variance was performed on the separate mean stroke volumes (4 samples of 5 items each; every item mean of 12 subjects). It appeared that there were no significant differences between the four samples (F = 1.9, P > 0.10). Study ZZZ.As shown in Table I, during a head-up tilt of 20° SI and CI were significantly lower than in the horizontal position (P G 0.05, Wilcoxon’s rank test for paired data), mainly as the consequence of a decrease in dZ/dt and T. Under these conditions MAP and TPR increased significantly (P G 0.05), whereas HR did not change. Study IV, The effects of blood-letting in two hypertensive patients (460 and 670 ml) are shown in Table 2. Although HR increased, CO decreased as a result of a decrease in SV. Despite a decrease in MAP, TPR in-

creased due to the large drop in CO. Figure 3 shows the effects. of blood withdrawal and reinfusion in a normal volunteer. During the withdrawal of 500 ml blood, MAP and HR increased slightly, while SV and CO decreased markedly. During reinfusion 2 days later, MAP and HR did not change, whereas SV and CO increased. The changes in SV were mainly the results of changes in dZ/ dt and Z0 (data not shown). Study V. Figure 4 shows the mean effect of closure of the artificial arteriovenous shunts of dialysis patients. For each patient, the means of 4-6 alternations between open and closed shunt states were taken. There were no effects in the patients with the small flow external shunts. In the patients with the large flow brachialis fistula shunts, HR, SI, and CI decreased significantly and MAP and TPR increased significantly. In the patients with the moderate flow radialis fistula shunts, these effects were less pronounced and significant. Study VI. Ultrafiltration in two dialysis patients resulted in losses of 1,600 and 1,200 ml ultrafiltrate (Table 3). Although HR increased, CO decreased as the result of a decrease in SV. Despite a decrease in MAP, calculated TPR increased due to the large drop in CO. Study VII. Table 4 gives the effects of volume depletion followed by 4 days of a Na-free diet in 20 patients with essential hypertension. This treatment caused a mean decrease of 0.5 1 in the plasma volume and of 2.0 1 in the extracellular fluid volume; the mean weight loss was 2.3 kg. HR increased significantly, but ST and CI decreased significantly (P s 0.001, Wilcoxon’s rank test for paired data). Study VIII. The effects of intravenous administration of propranolol and diazoxide on patients with essential hypertension are shown in Table 5, together with the CI simultaneously determined by thermodilution. According to both techniques, propranolol induced a significant decrease in HR and CI, whereas TPR increased; diazoxide caused a significant increase in HR, SI, and CI, but a decrease in TPR. A moderate degree of correlation (Spearman’s rank TABLE

1. Effect T

Supine 20” Tilt Supine

P

uf 20”

head-up tilting z,

MAP

HR

SI

CI

TPR

21.1 20.8 21.5

93 98 93 SO.05

65 63 61

76 59 73 SO.05

4.8 3.6 4.4 SO.05

810 1,140 890 SO.05

aat

0.303 0.278 0.304 SO.05

2.12 1.73 1.98 SO.05

Mean effect of 20” head-up tilting in 5 healthy volunteers on ventricular ejection time ( T), peak height (dZ/dt), thoracic impedance (&), mean arterial pressure (MAP), heart rate (HR), stroke index @I), cardiac index (Cl), and total peripheral resistance (TPR). Statistics: Wilcoxon rank test for paired data. TABLE

2. Effect of bloodletting T I

------MAP Control Bloodletting

154 140

Patient

A

l

1 Patient B (670 ml) ---- HR _ SV 7 !CO i TPR jMAP HI-i SV CO TPR I I 112 71 8.1 I 1,510 108 7s 144 11.1 770 123 56 6.9 1,670 93 t36 95 8.3 900 (4W

ml)

Effect of bloodletting in 2 patients. Volume removed in parentheses.

For abbreviations,

see Table

1.

Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (137.154.019.149) on January 10, 2019.

H494

BOER,

ROOS,

GEYSKES,

AND

DORHOUT

MEES

reinfusion

withdrawal

FIG. 3. Effect of withdrawal of 500 ml of blood reinfusion 2 days later in a healthy volunteer. abbrevations, see Table 1.

and For

t (min) A.

IlO-

M. AS? (mmHg)

B.

n=6

1

rm-

a-0

sLl

I-

j-

.-=

3 I-

I-

I-

800

HR. (min-I

)

I

70i

O-0

), -Is. I.

(ml.m-2)

I-

soLoI ?

l

-

0

I

T i

O-L..

-I-

TIL,

-I-

.