The Effect of Respiration on the Monitoring of Stroke Volume and Cardiac Output by the Electrical Impedance Technique

European Journal of Europ. J. Intensive Care Medicine 2, 3-6 (1976) Intensive Care Medicine © by Springer-Verlag 1976

J. Endresen and D. W. Hill

Research Department of Anaesthetics, The Royal College of Surgeons of England, London, England

Abstract. In six volunteers (5 male, 1 female) it has been shown that normal respiration made no statistical difference to the estimates of the mean stroke volume and the mean cardiac output as determined by the electrical impedance method of Kubicek et al, (1966). The coefficient of variation was usually increased by respiration. The use of those stroke volumes which occur only at end-expiration was not shown to yield a greater reproducibility with 3 other male volunteers. In the female subject it was found that the use of a digital averager triggered from the preceding R-wave of the ECG gave values for the mean stroke volume and cardiac output which were always lower than the conventional mean values obtained from a number of strokes. The expense of either of these approaches does not appear to be justified as a means of compensating for the effects of normal respiration on the impedance dZ/dt waveform. Key words: Impedance cardiogram, Cardiac output, Stroke volume, Averaging.

Introduction baseline, although the differentiator circuit makes the dZ/dt tracing less affected by respiration than the 2xZ The thoracic impedance technique developed by Kubicek tracing. et al. (1966) for the monitoring of changes in stroke volume and cardiac output has proved to be of value, for The movement of the dZ/dt tracing with respiration is example, in monitoring such changes occurring during an- most evident during automatic ventilation of the lungs aesthesia (Hill and Lowe, 1973; Lenz et al., 1976). How- with a tidal volume of 500 ml or more, and there is also ever, whereas the thoracic electrical impedance at 100 kHz the added complication that the phonocardiogram used may change by approximately 0.2 ohms with each systole for timing the duration of the left ventricular ejection is for an adult with a normal heart, a substantially larger often obliterated by the noise arising from the forceful change occurs with each tidal volume. A typical value inflation of the lungs. Under these circumstances, it is would be 2 ohms per litre (Baker and Hill, 1969). The advisable to stop the ventilator for a period of at least action of breathing can affect the waveform of both the six heart beats in order to obtain a representative length change in thoracic impedance (~Z) and that of its first of dZ/dt tracing. In many patients, the dZ/dt trace has a derivative (dZ/dt). The formula of Kubicek et al. (1966) well defined notch (the X-point) which occurs at the for the calculation of the stroke volume is closure of the aortic valve and then a phonocardiogram SVml = (pL2/Zoz) x T x (dZ/dt)max where "p" is the resis- is not vital for the measurement of T. For moderate auto- tivity of the patient's blood in ohm-cm, L is the distance matic ventilation or for normal spontaneous respiration in cm between the inner pair of the band electrodes it would be convenient not to have to stop the ventilator placed on the neck and lower portion of the chest, Z o is or to ask the subject to hold his breath each time a stroke the basal value of the thoracic impedance in ohms, T is volume estimate was made. Thus the objective of this the left-ventricular ejection time in seconds and study was to ascertain whether there was any statistical (dZ/dt)max is the peak value of the dZ/dt tracing obtain- difference between the mean stroke volume and the mean ing during systole in ohms per second. It is measured as cardiac output values estimated by the electrical impe- the peak height above the dZ/dt = 0 baseline. The action dance technique when the subject was breathing normal- of respiration causes the dZ/dt tracing to vary about this ly or breath-holding. 4 EuropeanJournal of Intensive Care Medicine, Vol. 2, No. 1 (1976)

Materials and Methods graph's phonocardiogram amplifier to provide a phonocardiogram. Finally, an ECG signal is obtained via the The subjects of this study consisted of eight males and outermost pair of the band electrodes. This is a non- one female aged 21 to 46 years. All were in good health standard diagnostic ECG lead, but it is adequate for tim- and had apparently normal hearts. For each subject two ing purposes and for triggering a digital averager. The disposable band electrodes made from self-adhesive Mylar impedance cardiograph provides 0.1 ohm and 1 ohm per strip carrying a 6 mm wide aluminium strip along the second calibration signals for the AZ and dZ/dt tracings. centre-line were placed around the neck spaced approx- The AZ, dZ/dt, ECG and phonocardiogram signals imately 25 mm apart. A second pair of band electrodes were displayed on an ink-jet Mingograf recorder together was placed around the bottom of the thorax with the with the output from a digital averager in the case of the inner electrode at the level of the xiphisternal joint, the female subject. The averager was a 200 point signal recov- electrode spacing again being 25 ram. The four electrodes ery system Type DL 102A (Data Laboratories Ltd.) were connected to a Model 304 Impedance Cardiograph which has 200 words of store. It was fed with the dZ/dt (Instrumentation for Medicine Inc.). A constant current signal and d.c. level triggered from the R-waves of the of approximately 4 mA r.m.s, at 100 kHz was passed ECG. between the outermost pair of the four electrodes (Fig. 1), Each subject lay supine, breathing normally and ade- and the potential changes due to cardiac activity detected quate time was allowed for the subject to relax and for at the inner pair of electrodes. The impedance cardio- the tracings to stabilise. A control sequence of at least graph back-off the potential developed due to the basal six beats was recorded together with the calibration. Im- impedance of the chest Z o and provides a signal output mediately afterwards the subject was requested to refrain corresponding to the change in thoracic impedance AZ. from breathing and a similar set of beats was recorded. The AZ signal is differentiated within the impedance Using the formula of Kubicek et aL (1966), the stroke cardiograph by a low-noise differentiator to provide a volume was calculated for each beat, the value of Z o hav- dZ/dt signal. A crystal microphone placed on the chest ing been obtained from the digital display of the imped- over the heart is connected to the impedance cardio- ance cardiograph. The instantaneous heart rate was cal-

~% Potential electrodes

OSC. f=100 kHz

d71~t Differentiator Current III / Amplifier J electrodes ][l/ and I "-/J ~ demodulator A Fig. 1. Diagramatic representation of the Impedance Cardiograph

L =24.3 cm Zo=32.252 (dZ) T=O.ZJ, sec ~ max : 1.5~2/sec

SV = 56.Z, ml HR=60 Q =3.38 [/min

Az=O.l~

dZ = l~/sec dt Baseline stabilizer unit output

I r I I I sec

Fig. 2. The dZ/dt signal "gated" using the IFM Baseline Stabiliser Unit J. Endresen and D. W. Hill: The Effect of Respgation on the Monitoring of Stroke Volume and Cardiac Output

Table 1. The cardiac output values are not corrected for haematocrit

Name Sex Mean Stroke Volume Mean Heart Rate Mean Cardiac Output

BL M (A) 152 ml (8.4%) 67.9 (2.9%) 10.3 L/min (6.2%) (B) 158 ml (14.0%) 65.6 (5.0%) 10.4 L/rain (16.0%) CO M (A) 128 ml (6.5%) 61.5 (2.5%) 7.8 L/min 4.8%) (B) 139 ml (11.7%) 61.3 (4.8%) 8.5 L/rain 14.4%) S-S M CA) 136 ml (5.5%) 54.7 (4.5%) 7.5 L/rain 5.3%) (B) i38 m! (5.0%) 53.9 (2.4%) 7.4 L/rain 7.4%) ME M (A) 105 ml (4.5%) 69.4 (3.6%) 7.4 L/rain (3.4%) (B) 99 ml (5.7%) 70.0 (2.6%) 6.9 L/rain 7.1%) BA M (A) 82 ml (2.2%) 64.0 (1.7%) 5.2 L/rain 3.0%) (B) 82 ml (4.8%) 68.0 (3.7%) 5.6 L/min 6.4%) EE F CA) 94 ml (5.3%) 76.0 (4.7%) 7.1 L/min 5.9%) (B) 84 ml (18.4%) 84.0 (4.6%) 7.0 L/rain (18.1%)

(A) Apnoeic, (B) Breathing, (%) Coefficient of variation.

Table 2.

Name Sex Mean Cardiac Output Averager Cardiac Output Difference Mean (R-(dZ/dt)max)

EE F (B) 7.51 L/min 7.04 L/min - 6.3% 96.4 ms (7.4%) EE F (B) 5.70 L/rain 5.68 L/min - 3.2% 96.4 ms (5.7%) EE F (B) 7.48 L/rain 7.08 L/rain - 5.4% 97.5 ms (7.8%) EE F (B) 6.67 L/rain 5.94 L/min -10.9% 93.1 ms (6.4%) EE F (B) 6.78 L/rain 6.31 L/rain - 7.2% 97.4 ms (6.3%) EE F (A) 7.80 L/rain 7.24 L/min - 7.2% 94.1 ms (5.7%) EE F CA) 6.98 L/rain 6.15 L/rain -11.9% 98.4 ms (4.7%)

(A) Apnoeic, (B) Breathing, (%) Coefficient of variation.

Table 3. Name Sex Mean of all Stroke Volumes Mean of end-expiratory Stroke Volumes

PO M (B) 119 ml (11.7%) 113 ml (6.7%) JE M (B) 111 ml (13.5%) 109 ml (10.4%) MO M (B) 157m1(9.4%) 161 ml(6.1%)

(B) Breathing, (%) Coefficient of Variation

culated for each beat from the R-R interval and the car- the IFM Baseline Stabiliser Unit where the AZ waveform diac output per beat determined by multiplying together is used as a gating signal (Fig. 2). A visual selection was the heart rate and the stroke volume values. For each set performed for three male subjects tracings over a period of readings the mean values were found together with of one hundred beats. the coefficients of variation and Student's t-test was employed to test for any significant difference between the corresponding breathing and apnoeic values. Results For the female subject, the average dZ/dt waveform provided by the digital averager from 4, 8, 16, 32 and 64 None of the six subjects tested showed any statistical beats was employed to determine the mean stroke volume difference at the p = 0.05 level between their mean stroke and this was compared with the average value obtained volume, mean heart rate or mean cardiac output values from all the readings in the individual batches by manual obtained when breathing normally or when breath hold- measurement. ing, (Table 1). For three of the subjects the coefficient of Since the AZ waveform is markedly affected by respi- variation for stroke volume and cardiac output was much ration, it can be used as a visual guide to the selection of greater during breathing than during breath-holding. For those stroke volumes which can occur close to end-expi- the other three there were smaller changes in the coeffi- ration. This selection can be performed automatically in cient of variation. When the digital averager was used the European Journal of Intensive Care Medicine, Vol. 2, No. 1 (1976) mean stroke-volume and cardiac output values obtained seems preferable simply to average the individual stroke for the female subject were consistently lower (Table 2), volume values and heart rates from at least two complete than those found by a conventional averaging of the respiratory cycles. series of individual values. Although there was no significant difference between the mean R4vave to (dZ/dt)max Acknowledgments. We are grateful to Group Captain A. J. Merri- intervals when breathing or breath-holding, the differ- fled for providing some of the tracings and to Albury Instruments Ltd. for the loan of the digital averager. The Vandervell Founda- ences between the individual values could explain the tion kindly provided the IFM Impedance Cardiograph and Base- reduction in the digitally averaged value of (dZ/dt)max. line Stabiliser Unit and the Medical Research Council provided Table 3 lists the mean stroke volume values and their computer facilities. One of us (J. E.) is indebted to the Wellcome variances for all the one hundred beats selected for three Trust for support. other male volunteers and the corresponding values for only those beats which occurred close to end-expiration. There was not any statistical difference between the References means. 1. Baker, L. E., Hill, D. W.: The use of electrical impedance technique for the monitoring of respiratory patterns during anaesthesia. Brit. J. Anaesth. 41, 2-17 (1969) Discussion 2. Hill, D. W., Lowe, H. J.: The use of the electrical impedance technique for the monitoring of cardiac output during The results of this study confirm that normal spontane- anaesthesia. Med. Biol. Engng. 11, 534-545 (1973) ous respiration does not significantly (p = 0.05) affect the 3. Kubicek, W. G., Karnegis, J. N., Patterson, R. P., Witsoe, estimates of mean stroke volume and mean cardiac out- D. A., Mattson, R. H.: Development and evaluation of an impe dance cardiac output system. Aerospace Med. 37, 1208- put obtained by the electrical impedance method of 1212 (1966) Kubicek et al. (1966). As would be expected, the effect 4. Lenz, R. J., Thomas, T. A., Wilkins, D. G.: Cardiovascular of respiration was to increase the variance of the values. changes during laparoscopy. Anaesthesia 31, 4 - 12 (1976) The use of an expensive digital averager to average out the effects of respiration is not justified. A cheaper approach to the reduction of the effect of respiration is J. Endresen D. W. Hill that of the IFM Baseline Stabiliser Unit, but although the Research Department of Anaesthetics variance of only those stroke volume values occurring The Royal College of Surgeons of England near end-expiration is less than that for all the beats, the Lincoln's Inn Fields mean values are not significantly different. In practice, it London WC2A 3PN, England