Comparison of Bioimpedance and Thermodilution Methods for Determining Cardiac Output: Experimental and Clinical Studies

Transcription

1 Comparison of Bioimpedane and Thermodilution Methods for Determining Cardia Output: Experimental and Clinial Studies Franis G. Spinale, M.S., Ph.D., H. David Reines, M.D., and Fred A. Crawford, Jr., M.D. ABSTRACT The hanges in eletrial bioimpedane aused by the blood flow through a thorai segment may be measured using a series of eletrodes plaed at opposing ends of this segment. Cardia output (CO) is alulated by omputer as the hange in bioimpedane over time. This study was performed to determine the auray of bioimpedane CO (CObi) ompared with standard thermodilution CO (COtd) in an animal model and in patients. COtd was determined using a thermodilution CO omputer, and CObi was alulated with a bioimpedane omputer in 1 dogs at baseline and at 5-minute intervals following the injetion of 7 meq of alium hloride. A high orrelation between COtd and CObi was observed (T =.91, p <.1) over a range of 2.5 to 6 L/min. Thirty-three paired observations using the CObi and COtd methods were performed in 1 patients following eletive oronary artery bypass surgery. A signifiant orrelation between the two methods was determined with a CO range of 2.1 to 7.8 L/min (T =.77, p <.1). CObi beame inaurate with severe tahyardia (pulse, 18/min), low CO, or frequent arrhythmias. These results indiate that this tehnique provides a ontinuous noninvasive method of determining CO that is omparable with thermodilution tehniques. The tehnique of CObi holds promise for widespread use when ontinuous and noninvasive monitoring of CO is neessary. The urrently aepted methods to estimate ardia output (CO) in linial pratie are Fik, dye-dilution tehniques, and thermodilution. The thermodilution method (COtd) for estimating CO has gained wide aeptane in ritial and oronary are units. COtd has been shown to be omparable to Fik and dye-dilution methods; however, signifiant errors in measurement have been doumented [l, 21. COtd also has the disadvantages of being invasive and providing only intermittent estimates of CO. An alternative method for estimating CO uses thorai eletrial impedane, or bioimpedane. Bioimpedane From the Department of Surgery, Medial University of South Carolina, Charleston, SC. Presented at the Thirty-Fourth Annual Meeting of the Southern Thorai Surgial Assoiation, Boa Raton, FL, Nov 57, Address reprint requests to Dr. Spinale, Division of Cardiothorai Surgery, Medial University of South Carolina, 171 Ashley Ave, Charleston, SC was first demonstrated to be of use in the late 196s for the measurement of ardiovasular performane in astronauts [3]. This method employs a low voltage similar to an eletroardiogram (1 mv), applied to the thorax, where hanges in the volume and veloity of the blood flow in the thorai aorta will result in detetable hanges in thorai ondutivity. This osillating omponent of the total thorai impedane (Z) an be expressed as its derivative (dz/dt). This derivative has been shown to be proportional to stroke volume (SV), and when the heart rate (HR) is known, an estimate of CO an be derived [4, 51. However, large disrepanies in CO in heterogeneous groups of patients with a wide spetrum of illnesses have been observed using this method [6]. More reently, improved sensitivity and refinements in the bioimpedane measurements through the use of miroproessor software have produed more aeptable results in estimations of CO in ritially ill patients [ These improvements in the measurement of CO using bioimpedane (CObi) have been programmed into the miroproessor of a ommerially available omputer alled the NCCOM 3 (BoMed Medial Manufaturing, Ltd., Irvine, CA). Eight thorai eletrodes allow measurement of HR, the base thorai impedane (thorai fluid index), the maximum rate in impedane hange (ejetion veloity index), and the ventriular ejetion time. With these parameters, the NCCOM 3 is able to alulate SV and CO (Fig 1). The purpose of this study was to ompare the appliability and auray of CObi with that of COtd in laboratory and linial settings. Materials and Methods Dogs Ten adult mongrel dogs weighing between 1 and 3 kg were used in the study and were ared for in aordane with the National Institutes of Health guidelines [12]. The dogs were anesthetized intravenously with sodium pentobarbital, 28 mg/kg, intubated, and plaed on a Harvard respirator at a tidal volume of 12 breathdmin. A 7F Swan-Ganz atheter was inserted through a femoral vein into the pulmonary artery to obtain COtd. COtd was determined by injetion of 1 ml of ied dextrose, and the thermal washout was alulated by omputer (Model SP1435; Gould In., Oxnard, CA); these values were obtained in tripliate. An eletroardiogram monitored HR, and a atheter plaed in the femoral artery monitored systemi blood pressure. 421 Ann Thora Surg 45: , Apr Copyright 1988 by The Soiety of Thorai Surgeons

2 422 The Annals of Thorai Surgery Vol45 No 4 April 1988 R x L2 = x VET x (dz/dt)max L where SV is Stroke Volume (ml) R is speifi resistivity of blood (OhmCm) L is thorai length (m) Z is thorai base (basal) impedane (ohm) (8z/dtlmax is the maximum rate of impedane hange during systoli upstroke (ohm/ se 1 Mehanial Ventilation t one heart beat 1 dz/dt [ ohm/se) h I\ (dz/dt)nax = EVI I p-4 one heart beat I Fig 1. The reorded hange in impedane over time an be equated to stroke volume (SV), as derived by Kubiek et a1 [31. (TFI = thorai fluid index; EVI = ejetion veloity index; VET = ventriular ejetion time.) (Reprodued with permission from BoMed Manufaturing Ltd, 1987.) Eight subutaneous needle eletrodes of the NCCOM 3 were plaed as speified by the manufaturer: Two of the four nek eletrodes were plaed at the base of the nek, and two were plaed 5 m ephalad. Two of the four thorai eletrodes were plaed level with the xiphoid, and two were plaed 5 m distally. Needle eletrodes were required beause of the diffiulty in plaing gel eletrodes on the animal's nek. Differenes between the geometri shape of the anine and human hest neessitated alibration of the bioimpedane omputer. Calibration was performed in 1 ontrol dog by using the mean value of 6 COtd determinations and adjusting the geometri length entered into the bioimpedane omputer until the CObi equaled the mean COtd. This length variable (L) is the measured distane between the inner set of eletrodes. Calibration in humans is not neessary sine orretion fators are preprogrammed into the software. Reordings inluded COtd and CObi measured in tripliate, mean arterial pressure, and HR. Following alibration and baseline measurements, 7 meq of alium hloride was infused into the proximal port of the venous atheter and mea- Fig 2. Experimental preparation showing position of leads and atheters. (COT, = thermodilution ardia output; BP = blood pressure; COB, = bioimpedane ardia output.) surements were performed 5, 1, and 15 minutes following infusion. The experimental design is illustrated in Figure 2. The data were subjeted to analysis of variane and regression analysis. Patients Simultaneous measurements of COtd and CObi were performed in randomly seleted patients following oronary artery bypass surgery. Eah patient had the prior plaement of a thermistor-tip pulmonary artery flowdireted atheter. All COtd values were alulated from the mean of at least three thermodilution injetions using 1 ml of ied dextrose. All injetions were made within 4 seonds without regard to the phalse of the respiratory yle. Eight eletroardiogram spot eletrodes were plaed aording to the manufaturer's instrutions, as shown in Figure 3. The geometri length variable (L) was entered diretly into the omputer as the distane between the inner set of eletrodes. The CObi measurements were taken as the mean value of 12 onseutive heartbeats. The patient data were analyzed using regression analysis and the Wiloxori test for paired observations.

3 423 Spinale et al: Thermodilution and Bioimpedane THERMODILUTION BIOIMPEDRNCE - 6 DOG #3 C f Y 4.5 Fig 3. The NCCOM 3 bioimpedane omputer and eletrodes attahed to patient E 6? m r 5 V a 2 3 u 5 2 F1 [L: a v CARDIAC OUTPUT-CObi (L/min) Fig 4. Sattergram showing omparison between bioimpedane ardia output (CObi) and thermodilution ardia output (COtd) measurements following alium hloride infusion in dogs. Limits defining 2% variability from the identity line are shown (r =.91, y =.78~ +.63, p <.1). Results Dogs Forty observations were obtained, four for eah dog. Calibration of the bioimpedane omputer to the determined COtd values in the ontrol dog resulted in an overall orretion oeffiient of.79x(l), where the mean thorai length for the animals in the study was 29 m. Regression analysis of CObi and COtd data resulted in an exellent fit (r =.91, p <.1; Fig 4). Individual response plots to alium hloride demonstrated that CObi refleted similar hanges as those obtained by BRSE I2 15 LINE TIME FOLLOWING CaCl INJECTION (min) Fig 5. Individual response urves for ardia output bioimpedane (CObi) and thermodilution (COtd) following alium hloride infusion in 2 representative dogs. CObi losely traked hanges in COtd values E I 3 5 a V E 3 n CARDIAC OUTPUT-Cob1 (L/mtn) Fig 6. Sattergram omparing bioimpedane ardia output (CObi) and thermodilution ardia output (COtd) measurements obtained from postoperative ardia surgery patients. Limits defining 2% variability from the identity line are shown (r =.77, y =.86~ +.74, p <.1). COtd over time (Fig 5). The range of COtd was from 2.4 to 6.2 L/min; the range of CObi was from 1.9 to 7.5 L/min. Patients The patient population inluded men and women ranging in age from 4 to 86 years. One patient was removed from the study beause a paemaker was required, resulting in erroneous CObi determinations. Thirtythree paired observations from 1 patients were obtained, with a range of 3.3 to 7.8 L/min for COtd and 2.1 to 7.8 L/min for CObi. There was no signifiant differene between the CO values obtained by thermodilution

4 424 The Annals of Thorai Surgery Vol45 No 4 April 1988 and by bioimpedane (p >.33). COtd and CObi were strongly related in this patient sample (r =.77, p <.1; Fig 6). Comment The eletrial impedane of the thorai avity hanges with blood flow [3, 41 beause of the inrease in arterial vessel size with pulsation and the alignment of red blood ells within the lumen with the inrease in blood veloity. These hanges inrease eletrial ondutane of the thorai avity and ause a bioimpedane waveform. The NCCOM 3 has been programmed to estimate SV and CO from this bioimpedane waveform with a known geometri length (L). In the animal study, beause of the variation in dog speies and the diffiulty in aurately determining the proper length variable, we alibrated CObi with COtd using a predetermined alibration fator (.79). Previous studies using this method of alibration have also demonstrated a high degree of orrelation between CObi and COtd in a variety of experimental onditions [13, 141. The primary question addressed by the animal study was whether CObi follows the trend of COtd after infusion of a positive inotropi agent. CObi followed losely COtd and was highly orrelated. Sine CObi estimates SV on a beat-to-beat basis and COtd is a mean value alulated from the thermodilution washout urve, this may aount for some of the variation observed between COtd and CObi. When the HR of the dogs exeeded 18 beats/min or when arrhythmias ourred, CObi beame inaurate and errati. This variation probably was due to the inability of the NCCOM 3 to detet the R wave of the eletroardiogram, resulting in an inorret SV determination. In the human study, CObi ould not be determined in 1 patient beause of the use of a paemaker. The urrent software installed in the NCCOM 3 was unable to differentiate the paemaker spike from the R wave, and extremely low CObi estimates were the result. The greatest perentage of variability enountered in the human study ourred at low flows (< 2.5 L/min), where CObi overestimated the respetive COtd alulation by 4 to 5% in 1 patient. Several explanations are possible for this observed overestimation: (1) An inorret value for L was entered into the omputer; (2) a omputer error was made in the alulation of dz/dt; (3) a ombination of user input and omputer errors ourred; and (4) the COtd determination was inorret. The length variable, L, is extremely sensitive, sine it is squared in the SV omputation (see Fig l), and small inrements of L will reflet large hanges in alulated CObi. In addition, poor thermodilution injetion tehnique or malfuntion of the thermodilution omputer, or both, may have aused some of the observed disparity [2]. Impedane CO has the distint advantage of being ontinuous and noninvasive. Patients do not have to be sedated or paralyzed for implementation of this devie, and the NCCOM 3 was not affeted by respiratory artifats. For this reason it has a muh wider potential usage in a larger group of patients who otherwise would not be onsidered andidates for pulmonary artery atheters. The ontinuous and noninvasive nature of the measurements makes this tehnology potentially very useful for titration of drug and fluid therapies as well as for long-term patient follow-up, neessary in suh instanes as ardia transplantation. The bioimpedane omputer used in this study provided a ontinuous readout of HR, SV, CO, and several other impedanederived parameters that ould be stored on omputer disk for trending and researh purposes. The bioimpedane omputer used in this study ould not reliably determine CO during tahyardia or with the use of paemakers. CObi overestimated COtd at low flows by a large margin, and ardia arrhythmias preluded aurate omputer measurement of CO due to poor eletroardiogram R-wave reognition neessary for omputing dudt. These findings suggest that modifiations in the software may be neessary before the NCCOM 3 an be of value in ritially ill patients during episodes of low CO and tahyardia. Reent studies have reported poor agreement with CObi and COtd in patients with severe sepsis and valvular insuffilieny [lo, 111. Beause of the ontinuous and noninvasive nature of impedane-derived measurements, this method potentially provides the ability to monitor a patient from admission to disharge as well as during subsequent return visits. Bioimpedane may also hold promise for the evaluation of pediatri patients, in whom a thermodilution atheter may be diffiult to use and repeated dextrose injetions would be ontraindiated. A reent study has demonstrated some suess using CObi in ritially ill hildren [ 151. Further studies are neessary to investigate the preision of bioimpedane ardiography in a wider range of ritially ill patients. Referenes 1. Van Grondelle A, Dithey RV, Groves BM, et al: Thermodilution method overestimates low ardia output in humans. Am J Physiol245:H69, Levett JM, Replogle RL: Thermodilution ardia output: a ritial analysis and review of the literature. J Surg Res 27392, Kubiek WG, Karnegis JN, Patterson RP, et al: Development and evaluation of an impedane ardia output system. Aerospae Med 1:28, Van de Water JM, Philips PA, Thouin LG, et al: Bioeletri impedane: new developments and linial appliation. Arh Surg 12:541, Kubiek WG, From AHL, Patterson RP, et al: Impedane ardiology as a noninvasive means to monitor ardia funtion. J Asso Adv Med Instrum 4:79, Van De Water JM, Miller IT, Milne EN, et al: Impedane plethysmography: a noninvasive means of monitoring the thorai surgery patient. J Thora Cardiovas Surg 6(5):641, Sramek BB: Noninvasive tehnique for measurement of

5 425 Spinale et al: Thermodilution and Bioimpedane ardia output by means of eletrial impedane. In: Proeed-ings of the Fifth International Conferene on Eletrial Bioimpedane, Tokyo, Japan, 1981, pp Bernstein D: Continuous noninvasive real-time monitoring of stroke volume and ardia output by thorai eletrial bioimpedane. Crit Care Med 14:898, Bernstein D: A new stroke volume equation for thorai eletrial bioimpedane: theory and rationale. Crit Care Med 14:94, Donovan KD, Dobb GJ, Woods WPD, Hokings BE: Comparison of transthorai eletrial impedane and thermodilution methods for measuring ardia output. Crit Care Med 14(12):138, Appel PL, Kram HB, Makabee J, et al: Comparison of measurements of ardia output by bioimpedane and thermodilution in severely ill surgial patients. Crit Care Med 14(11):933, National Institutes of Health: Guide for the Care and Use of Laboratory Animals. Revised edition. Bethesda, MD, National Institutes of Health (NIH publiation no ). 13. Tremper KK, Hufstedler SM, Barker SJ, et al: Continuous noninvasive estimation of ardia output by eletrial bioimpedane: an experimental study in dogs. Crit Care Med 14(3):231, Miles DS, Gotshall RW, Sexson WR: Evaluation of impedane ardiography in the anine pup. J Appl Physiol 6(1):26, MKinley DF, Pollak MM: A omparison of thorai bioimpedane to thermodilution ardia output in ritially ill hildren. Crit Care Med 15(4):358, 1987 Notie from the Southern Thorai Surgial Assoiation The thirty-fifth Annual Meeting of the Southern Thorai Surgial Assoiation will be held at the Maro Island Resort, Maro Island, FL, November 1-12, There will be a $125 registration fee for nonmember physiians exept for guest speakers, authors and oauthors on the program, and residents. There will be a $5 registration fee for attendees of the Postgraduate Course on Saturday, November 12, The Postgraduate Course of the Southern Thorai Surgial Assoiation has been expanded to a full day and will provide in-depth overage of thorai surgial topis seleted primarily as a means to enhae and broaden the knowledge of pratiing thorai and ardia surgeons. Members wishing to partiipate in the Sientifi Program should submit an original abstrat and one opy by May 15, 1988, to Robert M. Sade, M.D., Program Chairman, Southern Thorai Surgial Assoiation, 111 East Waker Dr, Chiago, IL 661. Abstrats must be sub- mitted on the Southern Thorai Surgial Assoiation abstrat submission form. These forms may be obtained from the Assoiation s offie or in this issue of The Annals of Thorai Surgery. All slides used during the presentation must be 35 mm. Manusripts of aepted papers must be submitted to The Annals of Thorai Surgery prior to the 1988 meeting or to the Seretary-Treasurer at the opening of the Sientifi Session. Appliations for membership should be ompleted by July 1,1988, and forwarded to the Southern Thorai Surgid Assoiation, 111 East Waker Dr, Chiago, IL 661. Gordon F. Murray, M.D. Seretary-Treasurer Southern Thorai Surgial Assoiation Basi Siene Center West Virginia University Medial Center Morgantown, WV 2656