- Email: [email protected]
Validation of bioimpedance estimates of cardiac output in preeclampsia
Validation of bioimpedance estimates of cardiac output in preeclampsia James A. Scardo, MD,a Janna Ellings, CNM,a Stephen T. Vermillion, MD,b and Suneet P. Chauhan, MDa Spartanburg and Charleston, South Carolina OBJECTIVE: We sought to evaluate the capacity of a new thoracic electric bioimpedance system to estimate cardiac output in patients with preeclampsia. STUDY DESIGN: We performed a prospective comparison of thoracic electric bioimpedance and echocardiographic M-mode volumetric estimates of cardiac output (in liters per minute) in preeclampsia. Subjects with preeclampsia who were chosen by means of strict criteria (either a systolic blood pressure ≥140 mm Hg or a diastolic blood pressure 90 mm Hg, or both, and proteinuria ≥300 mg in 24 hours or ≥+1 on repeat dipstick measurement 6 hours apart) were asked to participate in an institutional review board–approved study. Thoracic electric bioimpedance and echocardiography were performed with the patients in the left lateral recumbent position. Thoracic electric bioimpedance estimates were recorded at bedside; investigators were blinded to 3 simultaneously obtained echocardiographic M-mode estimates of cardiac output. Means were entered as the estimate for each patient. To satisfy the assumption of independent samples, only 1 estimate from each technique was used for each patient. Data were analyzed by Bland-Altman comparison. Hemodynamic and demographic variables are presented as mean ± SD. RESULTS: Fifteen patients were enrolled. Mean maternal age was 25.9 ± 4.8 years, and mean gestational age was 34.0 ± 3.5 weeks. Mean arterial pressure was 112 ± 14 mm Hg. There was good agreement of thoracic electric bioimpedance–derived and M-mode–derived cardiac output estimates. CONCLUSIONS: In patients with preeclampsia, thoracic electric bioimpedance estimates of cardiac output compare well with echocardiographic M-mode estimates. (Am J Obstet Gynecol 2000;183:911-3.)
Key words: Preeclampsia, hemodynamic monitoring, cardiac output
Cardiac output is the amount of blood (in liters) pumped by the heart per minute. Classically, it is defined as the product of heart rate and stroke volume. Cardiac output may be estimated invasively by means of either thermodilution or oxygen-extraction techniques. The thermodilution technique uses changes in temperature of an infusate from the right atrium to the pulmonary artery. This technique estimates right ventricular output, which is extrapolated clinically to estimate left ventricular output. The oxygen-extraction technique uses oximetry differences from the periphery and central circulation. This latter method is a more physiologic standard, but it assumes that there is a stable relationship between oxygen delivery and consumption in the patient under investigation. From the Spartanburg Regional Medical Centera and the Medical University of South Carolina.b Presented at the Twentieth Annual Meeting of the Society for MaternalFetal Medicine, Miami Beach, Florida, January 31–February 5, 2000. Reprint requests: James A. Scardo, MD, Director of Ob/Gyn Medical Education, Spartanburg Regional Medical Center, Suite 403, 853 N Church St, Spartanburg, SC 29303. Copyright © 2000 by Mosby, Inc. 0002-9378/2000 $12.00 + 0 6/6/108892 doi:10.1067/mob.2000.108892
Noninvasive methods to estimate cardiac output include ultrasonography and bioimpedance. Echocardiography uses volumetric estimates of stroke volume (either Doppler or M-mode estimates) and estimates of heart rate (either electrocardiographic or M-mode estimates) to calculate cardiac output. Alternatively, bioimpedance uses changes in electrical resistance to a lowvoltage current across the thorax with each heartbeat to calculate stroke volume. With each heartbeat, blood flowing down the aorta decreases the amount of resistance to this current and allows for an estimate of stroke volume. Heart rate is measured electrocardiographically and constitutes the product of heart rate and stroke volume. Assessment tools and techniques, such as echocardiography and bioimpedance, are desirable because of their noninvasive nature. Furthermore, the echocardiographic Doppler technique has been validated in preeclampsia.1 However, there is controversy regarding the ability of bioimpedance to be used to estimate cardiac output in patients with preeclampsia.2, 3 We report the only study to date that is a comparison of cardiac output estimates assessed by echocardiography and bioimpedance in patients with preeclampsia. 911
912 Scardo et al
October 2000 Am J Obstet Gynecol
Fig 1. Bland-Altman comparison of thoracic electric bioimpedence and M-mode echocardiographic estimates of cardiac output among preeclamptic patients. Diamonds, Differences.
Material and methods Patients with the confirmed diagnosis of preeclampsia were recruited for this investigational review board– approved study. Preeclampsia was defined by strict criteria requiring hypertension (either systolic blood pressure >139 mm Hg or diastolic blood pressure >89 mm Hg, or both) and proteinuria (>300 mg/d or >30 mg/dL [≥+1] on urine dipstick) sustained at least over a 6-hour interval of time. Magnesium sulfate infusion has been shown to increase cardiac output estimates in preeclampsia,4 and therefore all were receiving the same standard magnesium seizure prophylaxis protocol (4-g bolus and 2 g/h) and were at least 60 minutes into the continuous infusion. Patients were not considered candidates if they had preexisting chronic hypertension, had received antihypertensive therapy, were <18 years old, or were <24 weeks’ gestation. All patients received simultaneous bioimpedance (Cardiodynamics, San Diego, Calif) and M-mode echocardiographic (Toshiba America Medical Systems, Inc, Tustin, Calif) volumetric estimates of cardiac output while in left lateral recumbency (75°-90° to the bed, which was parallel with the floor). Furthermore, all patients were in this position for 15 minutes before hemodynamic assessment was begun. Echocardiographers, blinded as to the purpose of the study, were shielded by an opaque shield to hide the bioimpedance monitor. Stroke volume was estimated by a computer-generated conical reconstruction of left ventricle dimensions obtained from M-mode measurements, a standard echocardiogram software application provided by Toshiba America. A complete maternal echocardiogram was obtained after estimates of cardiac output. Estimates of cardiac output were generated auto-
matically as the average of 3 estimates and stored to disk in an operator-independent fashion at the bedside. Two similarly trained echocardiographers performed the echocardiography. A single investigator (James A. Scardo, MD) orchestrated the bioimpedance monitoring, which is considered operator independent. The 3 bioimpedance estimates correlating with the 3 echocardiographic estimates were recorded automatically to disk by the bioimpedance monitor software and averaged. Data were analyzed by the Bland-Altman method for assessing agreement between 2 methods of clinical measurement.5 Demographic variables were expressed as mean ± SD. Results Fifteen patients were enrolled. Mean maternal age was 25.9 ± 4.8 years, and mean gestational age was 34.0 ± 3.5 weeks. Ten of the patients were white, 3 were black, and 2 were Hispanic. Eleven were nulliparous. Four of the patients had thrombocytopenia (platelet count <100,000/ mm3). Five of the patients had proteinuria ≥3+ on urine dipstick measurement 6 hours apart. Two of the patients had ultrasonographic evidence of growth restriction and abnormal fetal umbilical and middle cerebral artery and ductus venosus Doppler findings. There was no identifiable hemodynamic pattern associated with demographic or clinical subgroups. A Bland-Altman measure of agreement analysis was performed and graphed (Fig 1). Comment Even before investigators in a National Institutes of Health multicenter trial of bioimpedance found excellent
Scardo et al 913
Volume 183, Number 4 Am J Obstet Gynecol
correlation with thermodilution-derived estimates of cardiac output in critically ill patients, bioimpedance estimates of cardiac output had been used widely in critical care units across the United States.6 One reason that the use of such monitors in obstetric patients has been limited to research settings is that validation results are mixed in published articles.2, 3, 7 However, software and hardware have gone through several generations of revision since the findings by Easterling et al3 of a poor correlation between bioimpedance-derived (NCCOM-3; BoMed, Irvine, Calif) and thermodilution-derived estimates of cardiac outputs in critically ill pregnant patients. By using a revised version (revision 6) of the NCCOM-3, Masaki et al2 found excellent correlation (r = 0.91; P < .001) between the 2 modalities in 9 critically ill peripartum patients. Additionally, in healthy pregnant patients at term, Clark et al7 found excellent correlation (r = 0.915) of bioimpedance (NCCOM-3, revision 7) and oxygen-extraction estimates of cardiac index. However, in that study bioimpedance estimates were found to be reliable only in lateral recumbent positions. That is to say, maternal positioning has been shown to be very important regarding reliability of cardiac output estimates. The NCCOM-3 monitors are no longer commercially available. The monitor in our study (Bio Z; Cardiodynamics, San Diego, Calif) has not been studied in patients with preeclampsia. Most of the measurements reported here are in the normal range for pregnancy. For this modality of hemodynamic assessment to be of additional clinical value, studies regarding validity in high and low cardiac output states, such as sepsis and cardiomyopathy, are needed. Our findings are relevant for several reasons. First, these data may allow for further revisions of software. The findings are also relevant in that they give further support to recent findings with bioimpedance that help define
the hemodynamic spectrum of preeclampsia,8 and these findings add support for this noninvasive technology in recent randomized clinical trials in preeclampsia.9 However, the most important aspect of this research is that it allows support for an operator-independent, noninvasive method of hemodynamic monitoring in preeclampsiarelated research. At this time we still believe that this technique should be used only in a research setting while awaiting further validative studies. REFERENCES
1. Easterling TR, Watts DH, Schmucker BC, Benedetti TJ. Measurement of cardiac output during pregnancy: validation of Doppler technique and clinical observations in preeclampsia. Obstet Gynecol 1987;69:845-50. 2. Masaki DI, Greenspoon JS, Ouzonian JG. Measurement of cardiac output in pregnancy by thoracic electrical bioimpedance and thermodilution. Am J Obstet Gynecol 1989;161:680-4. 3. Easterling TR, Benedetti TJ, Carlson KL, Watts DH. Measurement of cardiac output in pregnancy by thermodilution and impedance techniques. Br J Obstet Gynaecol 1990;96:67-9. 4. Scardo JA, Hogg BB, Newman RB. Favorable hemodynamic effects of magnesium sulfate in preeclampsia. Am J Obstet Gynecol 1995;173:1249-53. 5. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10. 6. Shoemaker WC, Wo CCJ, Bishop MH, Appel PL, Van de Water JM, Harrington GR, et al. Multicenter trial of a new thoracic electrical bioimpedance device for cardiac output estimation. Crit Care Med 1994;22:1907-12. 7. Clark SL, Southwick J, Pivarnik JM, Cotton DB, Hankins GDV, Phelan JP. A comparison of cardiac index in normal term pregnancy using thoracic electrical bioimpedance and oxygen extraction (Fick) techniques. Obstet Gynecol 1994;83:669-72. 8. Scardo JA, Kiser R, Dillon A, Brost BB, Newman RB. Hemodynamic comparison of mild and severe preeclampsia: concept of stroke systemic vascular resistance index. J Matern Fetal Med 1996;5:268-72. 9. Scardo JA, Vermillion ST, Newman RB, Chauhan SP, Hogg BB. A randomized, double blind, hemodynamic evaluation of nifedipine and labetalol in preeclamptic hypertensive emergencies. Am J Obstet Gynecol 1999;181:862-6.
Key words: Preeclampsia, hemodynamic monitoring, cardiac output
Cardiac output is the amount of blood (in liters) pumped by the heart per minute. Classically, it is defined as the product of heart rate and stroke volume. Cardiac output may be estimated invasively by means of either thermodilution or oxygen-extraction techniques. The thermodilution technique uses changes in temperature of an infusate from the right atrium to the pulmonary artery. This technique estimates right ventricular output, which is extrapolated clinically to estimate left ventricular output. The oxygen-extraction technique uses oximetry differences from the periphery and central circulation. This latter method is a more physiologic standard, but it assumes that there is a stable relationship between oxygen delivery and consumption in the patient under investigation. From the Spartanburg Regional Medical Centera and the Medical University of South Carolina.b Presented at the Twentieth Annual Meeting of the Society for MaternalFetal Medicine, Miami Beach, Florida, January 31–February 5, 2000. Reprint requests: James A. Scardo, MD, Director of Ob/Gyn Medical Education, Spartanburg Regional Medical Center, Suite 403, 853 N Church St, Spartanburg, SC 29303. Copyright © 2000 by Mosby, Inc. 0002-9378/2000 $12.00 + 0 6/6/108892 doi:10.1067/mob.2000.108892
Noninvasive methods to estimate cardiac output include ultrasonography and bioimpedance. Echocardiography uses volumetric estimates of stroke volume (either Doppler or M-mode estimates) and estimates of heart rate (either electrocardiographic or M-mode estimates) to calculate cardiac output. Alternatively, bioimpedance uses changes in electrical resistance to a lowvoltage current across the thorax with each heartbeat to calculate stroke volume. With each heartbeat, blood flowing down the aorta decreases the amount of resistance to this current and allows for an estimate of stroke volume. Heart rate is measured electrocardiographically and constitutes the product of heart rate and stroke volume. Assessment tools and techniques, such as echocardiography and bioimpedance, are desirable because of their noninvasive nature. Furthermore, the echocardiographic Doppler technique has been validated in preeclampsia.1 However, there is controversy regarding the ability of bioimpedance to be used to estimate cardiac output in patients with preeclampsia.2, 3 We report the only study to date that is a comparison of cardiac output estimates assessed by echocardiography and bioimpedance in patients with preeclampsia. 911
912 Scardo et al
October 2000 Am J Obstet Gynecol
Fig 1. Bland-Altman comparison of thoracic electric bioimpedence and M-mode echocardiographic estimates of cardiac output among preeclamptic patients. Diamonds, Differences.
Material and methods Patients with the confirmed diagnosis of preeclampsia were recruited for this investigational review board– approved study. Preeclampsia was defined by strict criteria requiring hypertension (either systolic blood pressure >139 mm Hg or diastolic blood pressure >89 mm Hg, or both) and proteinuria (>300 mg/d or >30 mg/dL [≥+1] on urine dipstick) sustained at least over a 6-hour interval of time. Magnesium sulfate infusion has been shown to increase cardiac output estimates in preeclampsia,4 and therefore all were receiving the same standard magnesium seizure prophylaxis protocol (4-g bolus and 2 g/h) and were at least 60 minutes into the continuous infusion. Patients were not considered candidates if they had preexisting chronic hypertension, had received antihypertensive therapy, were <18 years old, or were <24 weeks’ gestation. All patients received simultaneous bioimpedance (Cardiodynamics, San Diego, Calif) and M-mode echocardiographic (Toshiba America Medical Systems, Inc, Tustin, Calif) volumetric estimates of cardiac output while in left lateral recumbency (75°-90° to the bed, which was parallel with the floor). Furthermore, all patients were in this position for 15 minutes before hemodynamic assessment was begun. Echocardiographers, blinded as to the purpose of the study, were shielded by an opaque shield to hide the bioimpedance monitor. Stroke volume was estimated by a computer-generated conical reconstruction of left ventricle dimensions obtained from M-mode measurements, a standard echocardiogram software application provided by Toshiba America. A complete maternal echocardiogram was obtained after estimates of cardiac output. Estimates of cardiac output were generated auto-
matically as the average of 3 estimates and stored to disk in an operator-independent fashion at the bedside. Two similarly trained echocardiographers performed the echocardiography. A single investigator (James A. Scardo, MD) orchestrated the bioimpedance monitoring, which is considered operator independent. The 3 bioimpedance estimates correlating with the 3 echocardiographic estimates were recorded automatically to disk by the bioimpedance monitor software and averaged. Data were analyzed by the Bland-Altman method for assessing agreement between 2 methods of clinical measurement.5 Demographic variables were expressed as mean ± SD. Results Fifteen patients were enrolled. Mean maternal age was 25.9 ± 4.8 years, and mean gestational age was 34.0 ± 3.5 weeks. Ten of the patients were white, 3 were black, and 2 were Hispanic. Eleven were nulliparous. Four of the patients had thrombocytopenia (platelet count <100,000/ mm3). Five of the patients had proteinuria ≥3+ on urine dipstick measurement 6 hours apart. Two of the patients had ultrasonographic evidence of growth restriction and abnormal fetal umbilical and middle cerebral artery and ductus venosus Doppler findings. There was no identifiable hemodynamic pattern associated with demographic or clinical subgroups. A Bland-Altman measure of agreement analysis was performed and graphed (Fig 1). Comment Even before investigators in a National Institutes of Health multicenter trial of bioimpedance found excellent
Scardo et al 913
Volume 183, Number 4 Am J Obstet Gynecol
correlation with thermodilution-derived estimates of cardiac output in critically ill patients, bioimpedance estimates of cardiac output had been used widely in critical care units across the United States.6 One reason that the use of such monitors in obstetric patients has been limited to research settings is that validation results are mixed in published articles.2, 3, 7 However, software and hardware have gone through several generations of revision since the findings by Easterling et al3 of a poor correlation between bioimpedance-derived (NCCOM-3; BoMed, Irvine, Calif) and thermodilution-derived estimates of cardiac outputs in critically ill pregnant patients. By using a revised version (revision 6) of the NCCOM-3, Masaki et al2 found excellent correlation (r = 0.91; P < .001) between the 2 modalities in 9 critically ill peripartum patients. Additionally, in healthy pregnant patients at term, Clark et al7 found excellent correlation (r = 0.915) of bioimpedance (NCCOM-3, revision 7) and oxygen-extraction estimates of cardiac index. However, in that study bioimpedance estimates were found to be reliable only in lateral recumbent positions. That is to say, maternal positioning has been shown to be very important regarding reliability of cardiac output estimates. The NCCOM-3 monitors are no longer commercially available. The monitor in our study (Bio Z; Cardiodynamics, San Diego, Calif) has not been studied in patients with preeclampsia. Most of the measurements reported here are in the normal range for pregnancy. For this modality of hemodynamic assessment to be of additional clinical value, studies regarding validity in high and low cardiac output states, such as sepsis and cardiomyopathy, are needed. Our findings are relevant for several reasons. First, these data may allow for further revisions of software. The findings are also relevant in that they give further support to recent findings with bioimpedance that help define
the hemodynamic spectrum of preeclampsia,8 and these findings add support for this noninvasive technology in recent randomized clinical trials in preeclampsia.9 However, the most important aspect of this research is that it allows support for an operator-independent, noninvasive method of hemodynamic monitoring in preeclampsiarelated research. At this time we still believe that this technique should be used only in a research setting while awaiting further validative studies. REFERENCES
1. Easterling TR, Watts DH, Schmucker BC, Benedetti TJ. Measurement of cardiac output during pregnancy: validation of Doppler technique and clinical observations in preeclampsia. Obstet Gynecol 1987;69:845-50. 2. Masaki DI, Greenspoon JS, Ouzonian JG. Measurement of cardiac output in pregnancy by thoracic electrical bioimpedance and thermodilution. Am J Obstet Gynecol 1989;161:680-4. 3. Easterling TR, Benedetti TJ, Carlson KL, Watts DH. Measurement of cardiac output in pregnancy by thermodilution and impedance techniques. Br J Obstet Gynaecol 1990;96:67-9. 4. Scardo JA, Hogg BB, Newman RB. Favorable hemodynamic effects of magnesium sulfate in preeclampsia. Am J Obstet Gynecol 1995;173:1249-53. 5. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10. 6. Shoemaker WC, Wo CCJ, Bishop MH, Appel PL, Van de Water JM, Harrington GR, et al. Multicenter trial of a new thoracic electrical bioimpedance device for cardiac output estimation. Crit Care Med 1994;22:1907-12. 7. Clark SL, Southwick J, Pivarnik JM, Cotton DB, Hankins GDV, Phelan JP. A comparison of cardiac index in normal term pregnancy using thoracic electrical bioimpedance and oxygen extraction (Fick) techniques. Obstet Gynecol 1994;83:669-72. 8. Scardo JA, Kiser R, Dillon A, Brost BB, Newman RB. Hemodynamic comparison of mild and severe preeclampsia: concept of stroke systemic vascular resistance index. J Matern Fetal Med 1996;5:268-72. 9. Scardo JA, Vermillion ST, Newman RB, Chauhan SP, Hogg BB. A randomized, double blind, hemodynamic evaluation of nifedipine and labetalol in preeclamptic hypertensive emergencies. Am J Obstet Gynecol 1999;181:862-6.