Hemodynamic monitoring in critically ill veterinary subjects continues to become more sophisticated with standards of care modeled after examination of critically ill humans. Cardiac output (CO) is the best available variable to assess overall cardiovascular function. Measurement of CO, along with blood hemoglobin concentration and oxygen saturation of hemoglobin, allows calculation of global tissue oxygen delivery and consumption, thereby providing useful information in individuals with primary cardiac disease or secondary cardiovascular derangements associated with systemic illness. Following trends in CO in individual animals in the intensive care unit might allow for both earlier detection of cardiovascular derangements and optimization of clinical interventions. To date, CO monitoring in adult horses has been limited to the research setting.[2-4] Cardiac or peripheral artery catheterization is required for indicator dilution methods of CO measurement and is generally not suitable in a clinical setting. Transthoracic echocardiography has been used to measure CO in people and in small animals with various 2-dimensional (2-D) volumetric or Doppler methods.[5, 6] Ultrasonography is widely available in equine hospitals, and most ultrasound units have software packages allowing calculation of various cardiovascular parameters including CO. Therefore, transthoracic echocardiography may represent a convenient and noninvasive means of measuring CO in equine patients. In one study, Doppler echocardiographic measurement of CO was found to agree well with the thermodilution method in adult horses. However, difficulties in aligning the ultrasound beam parallel to the blood flow and individual variability in the cardiac window can make this method difficult to use. In addition, indices of cardiac function derived from Doppler echocardiography have been found to be less repeatable than indices derived from 2-D or M-mode echocardiography in horses. It was recently shown that some volumetric echocardiographic methods provide an accurate and noninvasive estimate of CO in anesthetized neonatal foals. However, because of major differences in cardiac chamber sizes and inability to obtain apical views of the heart in adult horses, data generated from neonatal foals under general anesthesia cannot be directly extrapolated to adult horses.
The purpose of this study was to assess and validate various transthoracic echocardiographic methods of measuring CO in standing adult horses over a range of CO by comparing results to the lithium dilution CO (LiDCO) method. The hypothesis of the study reported herein was that volumetric methods would have better agreement with lithium dilution than Doppler-based methods.
- Top of page
- Materials and Methods
Twenty-four pairs of lithium dilution/echocardiography CO measurements were taken from the 8 horses. Adverse effects in this study ranged from mild to moderate hematoma formation at the site of arterial catheterization to transient collapse secondary to a suspected air embolus in 1 horse. Adverse reactions attributable to the dobutamine infusion included irritability and kicking out violently without cause in 2 horses, synchronous diaphragmatic flutter in 3 horses, and intermittent ventricular premature contractions in 1 horse.
Lithium dilution measurements of CO ranged between 16.6 and 63.0 L/min (mean ± SD = 30.5 ± 9.5 L/min), resulting in cardiac indices ranging between 17.5 and 129.6 mL/kg/min (59.9 ± 21.5 mL/kg/min). Lithium dilution determinations of CO during administration of dobutamine (37.7 ± 11.7 L/min) were significantly higher than determinations obtained at baseline (28.6 ± 4.7 L/min) or during administration of romifidine (25.1 ± 6.3 L/min). Cardiac output during administration of romifidine was not significantly different from CO obtained at baseline. Heart rates during administration of dobutamine (37 ± 6 beats/min) were significantly higher than heart rates obtained at baseline (32 ± 2 beats/min) or during administration of romifidine (29 ± 5 beats/min). Heart rate during administration of romifidine was not significantly different from heart rate obtained at baseline.
The analysis of bias indicated a significant effect (P < .001) of method of CO measurement, but no significant effect of level (low, intermediate, or high) of CO (P = .938) on bias, indicating that the performance of each echocardiographic method was not influenced by the magnitude of CO. The mean bias and limits of agreements for each echocardiographic method of CO measurement are presented in Table 1. The overall performance of each method was assessed by comparing the absolute value of their bias. Analysis of variance revealed a significant effect of method of CO measurement (P < .001), but no significant effect of level of CO (P = .089) and no significant interactions between level and method (P = .607). The absolute values of the bias of the 4C AL, 4C MOD, RVOT Doppler, and Bullet methods were significantly lower than that of LVOT Doppler or cubic methods (Table 1). Bland-Altman plots for the 4C AL, 4C MOD, RVOT Doppler, and Bullet methods are presented in Figure 1. The mean CO (±SD) for each method of measurement at each level of CO is reported in Table 2.
Table 1. Summary statistics of the agreement between cardiac output measurements (L/min) by lithium dilution and various echocardiographic methods
|Cardiac Output Methods||Bias (L/min ± s.d.)a||Limits of Agreement (L/min)||Absolute Value of Biasb(L/min)|
|Doppler LVOT||−14.2 ± 10.3||−34.3 to 5.9||14.8b|
|Doppler RVOT||2.2 ± 7.9||−13.4 to 17.7||5.5a|
|4-chamber area-length||−1.1 ± 8.4||−17.6 to 15.4||6.5a|
|4-chamber modified Simpson||0.8 ± 7.8||−14.4 to 16.0||6.1a|
|Bullet||−6.9 ± 9.2||−25.1 to 11.2||8.8a|
|Teichholz||7.6 ± 7.14||−6.5 to 21.6||10.4a,b|
|Cubic||−24.3 ± 18.0||−59.6 to 11.0||24.3b|
Table 2. Mean cardiac output (L/min ± SD) as determined concurrently by lithium dilution and various Doppler or volumetric echocardiographic methods. Measurements were obtained at baseline, after administration of dobutamine (high CO) and after administration of romifidine (low CO)
|LiDCO||28.7 ± 4.7||36.6 ± 10.0||25.1 ± 6.3|
|Doppler LVOT||46.0 ± 12.1||52.7 ± 9.6||35.4 ± 8.1|
|Doppler RVOT||29.0 ± 5.6||30.4 ± 4.4||26.1 ± 6.6|
|4C AL||30.0 ± 9.7||39.8 ± 10.8||25.6 ± 7.1|
|4C MOD||28.3 ± 8.8||37.3 ± 9.9||24.9 ± 3.3|
|Bullet||34.6 ± 7.1||46.2 ± 11.3||23.4 ± 3.8|
|Teichholz||20.5 ± 6.0||28.5 ± 8.4||19.8 ± 6.5|
|Cubic||47.6 ± 17.7||67.9 ± 22.3||48.9 ± 22.0|
Figure 1. Bland-Altman plots of CO values measured concurrently by lithium dilution and echocardiography by the 4-chamber modified Simpson (A), 4-chamber area-length (B), Doppler RVOT (C), or Bullet (D) methods in standing adult horses. The solid line represents the mean bias and the dashed lines represent the upper and lower limits of agreements (1.96 × SD). Three measurements were obtained from each of 8 horses for a total of 24 observations. Each symbol represents an individual horse.
Download figure to PowerPoint
- Top of page
- Materials and Methods
In this study, the 4C AL, 4C MOD, Doppler of the RVOT, and Bullet methods of cardiac output measurement in adult standing horses had better agreement with the lithium dilution method than the other methods evaluated herein. Ideally, these noninvasive echocardiographic methods of CO estimation would allow clinicians to monitor trends in CO, which would complement current hemodynamic monitoring tools available for use in critically ill adult horses. Cardiac output measurement is routinely employed in human critical care facilities and has become a common modality for hemodynamic monitoring in foals in some neonatal intensive care units. Currently, validated and accepted methods for estimation of cardiac output in adult horses are not suitable for routine clinical use as they are invasive, requiring maintenance of a peripheral arterial catheter or a pulmonary artery catheter for lithium dilution or thermodilution techniques, respectively.
The 4C AL and 4C MOD methods overestimated and underestimated CO, respectively, each by a mean of approximately 1 L/min, whereas the Bullet method overestimated the CO by approximately 7 L/min, relative to the LiDCO reference. When considering performance of the Doppler methods, agreement between the RVOT Doppler and the LiDCO was not significantly different from the best volumetric measures of CO, disproving the hypothesis of this study that the volumetric measures of CO would be superior to the Doppler methods. The RVOT Doppler method underestimated CO by a mean of 2 L/min relative to LiDCO, whereas the LVOT Doppler method overestimated CO by a mean of 14 L/min relative to LiDCO.
Although the 4C AL, 4C MOD, and RVOT Doppler methods had a significantly lower bias than the other methods evaluated, their limits of agreement with LiDCO were wide (±17 L/min), representing up to ±50% of the measured CO. Two methods of measurement can usually be considered interchangeable if the limits of agreement fall within a clinically acceptable range. A meta-analysis of studies using bias and precision statistics to compare CO measurement techniques in people underscored that considerable diversity exists in how the results of bias and precision are interpreted between studies. In the aforementioned study, the authors proposed that acceptance of a new technique for measurement of CO in people should rely on limits of agreement up to ±30%. The limits of agreement that would be considered acceptable for the measurement of CO in a horse have not been defined and might even vary depending on the clinical situation. Despite the wide limits of agreement, 4C AL, 4C MOD, and RVOT Doppler might prove to be of value to detect changes in magnitude and direction of CO in a clinical setting.
The carotid artery was catheterized owing to difficulties in maintaining patency in smaller peripheral arteries in conscious standing horses. An experimental study by one of the current authors reported maintenance of a carotid artery catheter for up to 24 hours without adverse effects. Although placement of a carotid artery catheter under ultrasound guidance was not technically challenging, the authors do not recommend carotid artery catheterization in client-owned horses. The use of pulmonary artery catheters has been reported to increase mortality and predispose to pulmonary thromboembolism in people. In addition, pulmonary artery catheterization has been associated with traumatic endocardial lesions in the right heart and pulmonary artery of adult horses. Thus, to eliminate the need for arterial catheterization in a clinic setting, the main purpose of this study was to determine which echocardiographic measures of CO most robustly correlate with an indicator dilution technique in standing horses.
Excellent image quality is essential to be able to make accurate and repeatable echocardiographic measurements. Poor image quality caused by subject factors (poor compliance, body condition score, differences in the cardiac window) was encountered to some extent in this study. This may have influenced the performance of some of the volumetric methods that required tracing the blood endocardium interface (Bullet, 4C AL, 4C MOD). However, the Teichholz and Cubic methods derived from standard M-mode images and left ventricular study measurements had poor agreement with the LiDCO method, significantly underestimating and overestimating the CO, respectively. Poor agreement was attributed to geometric assumptions that rely on a smooth circular or elliptical-shaped ventricle and uniform contraction, which may not be applicable to the equine left ventricle.[29, 30] Results from the Teichholz and Cubic-based methods were similar to results in anesthetized foals.
In this study, the LVOT Doppler method in the left parasternal window significantly overestimated the LiDCO reference. This is in contrast to results of a previous study comparing Doppler echocardiographic methods in standing adult horses in which LVOT Doppler in the left parasternal window had better agreement with thermodilution than RVOT Doppler. The discrepancy between the aforementioned and this study may be attributable to differences in the population of horses studied or poor repeatability of pulsed-wave Doppler measures in general.[7, 31, 32] Pulsed-wave Doppler measurement of aortic VTI was found to have more day-to-day and intercardiac cycle variation than pulsed-wave Doppler measurement of pulmonary artery VTI in one equine study. Although it is possible that the aortic diameter was overestimated, this was felt to be unlikely as measurements were made at the sinotubular junction and were within normal reference ranges for adult standing horses.[32-34] It is more likely that the pulsed-wave Doppler traces were too broad owing to artifact from poor alignment with blood flow. Overestimation of CO by pulsed-wave Doppler methods has been previously reported in both human and animal studies.[10, 32, 35, 36]
Inherent limitations exist with all CO-monitoring methods available for clinical use. Although there is no universally accepted gold standard for CO determination in human or veterinary medicine, thermodilution has traditionally been the most commonly used method in people.[18, 37] Studies comparing LiDCO to thermodilution CO have been performed in several species including the horse.[38-42] As a result of the relatively narrow limits of agreement between the 2 methods, the use of the LiDCO method as the reference standard method is widely accepted across many species.[39-43] Potential sources of error with the lithium dilution method include intracardiac shunts, which were ruled out by our inclusion criteria, and lithium accumulation. Lithium accumulation creates background “noise” and can result in overestimation of cardiac output. Accumulation was unlikely in this study as the mean cumulative lithium dose was 0.038 mmol/kg and no more than 8 lithium injections were administered to any horse (19.2 mmol LiCl total). Considerably higher cumulative dosages of LiCl (0.8 mmol/kg; 69.3 mmol LiCL) were used in a study in exercising horses and did not result in any adverse effects; however, overestimation of CO was documented as the number of lithium injections and exercise intensity increased.
Lithium determinations of CO in this study ranged between 16.6 and 63.0 L/min, which represent a wide physiological range of CO and are similar to ranges achieved in comparable studies.[7, 44] Echocardiographic derivation of CO was not influenced by level of CO. Although a statistically significant decrease in CO from baseline was not achieved with the romifidine CRI, our purpose of measuring CO over a wide range of values was met.
Limitations of the study design included a small and diverse sample population in terms of breed. Although the inclusion of various breeds may be viewed as a limitation and indeed may have impacted the performance of some of the transthoracic echocardiographic methods, the use of a monomorphic sample would be unrepresentative of most clinic populations, making it difficult to extrapolate the findings of this study to the target population. Another potential limitation included the inability to simultaneously measure CO with lithium dilution and echocardiography. Cardiac output is dynamic with beat-to-beat variation based on neuroendocrine input. Agreement between the methods might have been influenced by making measurements at discrete time points. Efforts to minimize this effect were made by performing lithium dilution measurements just before and immediately after the storage of transthoracic echocardiographic images and by performing all measurements stall side in a quiet environment to minimize transient excitation during measurements.
A potential source of error and limitation of transthoracic echocardiography is that the Doppler-derived techniques and some of the 2-D techniques require measurements derived by tracing velocity spectra or blood-endocardium interfaces, which are subject to observer interpretation. In the study herein, only 1 person derived measurements from the stored video clips, thus interobserver agreement was not evaluated. Other studies will be necessary to assess the reproducibility of these methods and to determine the effect of the observer on variability.
In the hands of adequately trained clinicians, transthoracic echocardiography of critically ill adult horses has advantages beyond CO measurement. Previous work has shown that volume depletion can be recognized rapidly based on chamber morphology. In addition, detection of diastolic or systolic dysfunction, valvular regurgitation, chamber dilation, regional wall abnormalities, or pericardial effusion would impact therapeutic decisions. The disadvantages of transthoracic echocardiography pertain to initial capital costs of purchasing an ultrasound unit and acquiring the training necessary to appropriately apply these techniques and interpret the results.
In conclusion, transthoracic echocardiography by the 4-chamber area-length method, 4-chamber modified Simpson method, Bullet method, or Doppler of the RVOT have significantly lower biases than all other methods evaluated in standing healthy adult horses. These methods of estimating cardiac output provide a noninvasive clinically accessible method for serial hemodynamic monitoring and warrant further investigation in critically ill adult horses.