Introduction
Transthoracic echocardiography (TTE) has become the leading noninvasive examination to assess the hemodynamic status of patients hospitalized in intensive care units (ICUs) [
1‐
3]. It allows the physician to study noninvasively the contractility, relaxation, and fluid responsiveness of the two ventricles [
4]. All parameters must be recorded in one of the following three windows: parasternal, apical, or subxiphoid. As recommended in guidelines, the appropriate window corresponds to the Doppler mode with the best alignment obtained between the ultrasound beam and the direction of blood flow [
5]. For that reason, most parameters recorded for hemodynamic assessment classically require an apical window: left ventricular ejection fraction (LVEF) [
6,
7], ratio between early and late mitral inflow (E/A) [
8], ratio between early mitral inflow and displacement of the mitral annulus on tissue Doppler imaging (E/Ea) [
8], velocity time integral of aortic blood flow (Ao VTI) [
9], and the ratio between right (RVEDA) and left (LVEDA) ventricular end-diastolic area [
7]. However, many factors in the ICU can interfere with apical image quality. More specifically, in this setting, the lung interposition between the probe and the heart, the presence of mechanical ventilation, the inability to position the patient in left-sided decubitus, tachypnea, and the presence of bandages or tubes on the chest can alter the quality of the images obtained on apical and parasternal views.
To complete the examination in this setting, the echocardiographer may perform transesophageal echocardiography. However, transesophageal echocardiography is more invasive and can be difficult to perform, especially in nonintubated ICU patients. In this setting, use of the transthoracic subxiphoid view can be an attractive alternative to transesophageal echocardiography for less-experienced operators who may be reluctant to perform an invasive examination. The subcostal view is classically used to study respiratory variations of the inferior vena cava and the presence of pericardial effusion. This view can provide additional hemodynamic parameters, such as LVEF, mitral or aortic flows, and RV/LV areas. Although the subxiphoid view does not provide the same alignment as the apical view, it is not known to what degree this different alignment affects the results of the hemodynamic examination.
To address this issue, we tried to determine the accuracy of the subxiphoid window to record various echocardiographic parameters usually recorded on the apical view for hemodynamic assessment.
Materials and methods
Patients
In accordance with French legislation, the local institutional review board (CPP Nord-Ouest II, Amiens University Hospital, France) approved the study protocol, and all patients or their relatives gave their informed consent. This single-center prospective study was performed in an eight-bed teaching hospital medical ICU (Amiens, France). Over a 4-month period, all patients admitted to the ICU were included in the study, except when one of the two experienced echocardiographers (JM and AS) was absent and except for patients admitted to the unit for less than 24 hours. Patient characteristics were recorded.
Measurements and analyses
Transthoracic echocardiography was performed during the first 2 days after admission in all patients by one of the two operators (JM or AS) by using a Vivid i (General Electrics Healthcare Fairfield, CT, USA) equipped with a 3-MHz probe. Parasternal (long and short axis), apical four-chamber, and subxiphoid four-chamber views were obtained successively in 2D mode, and a cine-loop was stored for offline analysis. In the apical and subxiphoid windows, pulsed Doppler mitral inflow was recorded with the sample cursor placed at the mitral valve tips; early diastolic velocity (Ea) was determined from Doppler tissue imaging with the sample cursor placed in the lateral mitral annulus, and aortic blood flow was measured with the sample volume placed at the aortic annulus. The size of the sample volume was not modified between apical and subxiphoid views.
The two operators then blindly reviewed the quality of the 2D images recorded with these windows offline on a computer (General Electric Healthcare, Fairfield, CT, USA), according to the following scoring system: Excellent (visualization of the four chambers with complete endocardial visualization of both ventricles); Good (visualization of the two ventricles with complete endocardial visualization of both ventricles); Acceptable (visualization of both ventricles, but short segments of endocardium are not correctly visualized); Poor (the endocardium cannot be visualized in most portions of the ventricles); and No image (no visualization of the ventricles). An adequate acoustic window was defined by an acceptable, good, or excellent score. Each physician separately scored the video recording of the three windows (parasternal, apical, and subxiphoid) and compared their results. When both scores were identical, the final score was accepted. When the two scores were different, the physicians jointly visualized the video, discussed their points of view, and reached a consensus to establish a final score.
In the subpopulation of patients with adequate apical and subxiphoid windows (scored excellent, good, or acceptable), LVEF (according to the Monoplane Simpson method) and RVEDA/LVEDA ratio were measured on both the apical and subxiphoid 2D cine-loops. The mitral peak E velocity (E), peak A velocity (A), E/A ratio, Ao VTI, and E/Ea ratio were measured on Doppler images in the apical and subxiphoid windows. All data were the average of three end-expiratory measurements (excluding outliers). In the presence of an outlier, an additional assessment was recorded. In the subxiphoid window, LVEF and RVEDA/LVEDA were recorded on a longitudinal four-chamber view visualizing both right and left ventricles in their largest dimensions. On a modified four-chamber view, the operator tried to obtain the best alignment between flow and the Doppler axis by obtaining the highest maximum Doppler velocity to record mitral and aortic flows on the subxiphoid view.
Statistics
All variables are expressed as mean ± 1 SD or proportions. The normal distribution was tested with a Kolmogorov-Smirnov test. Quantitative variables were compared by using a Student t test, and proportions were compared with a χ2 test.
Pearson correlation coefficient and Bland-Altman analysis were used to assess the adequacy of the various parameters obtained in the subxiphoid and apical windows. This analysis was performed in the subpopulation of patients in whom apical and subxiphoid windows were scored as excellent, good, or acceptable (
n = 65). The bias and limit of agreement (LOA) around the bias were given by the Bland-Altman representation. However, no consensus has been reached concerning the acceptable LOA to confirm validation of the subcostal view. To define an acceptable LOA, Cecconi
et al. [
10] proposed assessing the precision of the reference technique (in this case, the apical view) and the tested technique (the subxiphoid view). We prospectively defined the validation criteria for the various parameters studied on the subxiphoid view: to be acceptable, the parameter on the subxiphoid view had to be accurate (bias <10%) and precise (a lower percentage precision value than on the apical view).
The inter- and intraobserver reproducibilities were determined for each parameter (LVEF, E/A, E/Ea, RVEDA/LVEDA, and Ao VTI) in both the apical and subxiphoid windows. All intraobserver variabilities were <5%, and interobserver variabilities were <9%.
Statistical analysis was performed by using MedCalc version 12.0.4.0 (MedCalc Software, Mariakerke, Belgium). The limit of significance was set at ≤ 0.05.
Discussion
In this study, the apical view was the view most frequently obtained (86%), and only a small proportion (5%) of patients did not have an adequate apical view, but had an adequate subxiphoid view. These results emphasize the good feasibility of the apical view in TTE, even in ventilated or COPD patients. This study also shows that the subxiphoid view is a reliable view to study the right- and left-ventricular morphology and function on two-dimensional echocardiography, but not in Doppler mode. The subxiphoid view should be used to explore only the vena cava, pericardium, and the morphology of the left or right ventricles.
The useful window most frequently obtained in our population was the apical window (86%), followed by the parasternal window (66%), and finally the subxiphoid window (49%), as already reported [
11‐
13]. Only 5% of the patients of our population presented an inadequate apical view and an adequate subxiphoid view. Moreover, a greater number of parameters were recorded on the apical view than on the subxiphoid view. Altogether, these results emphasize the better feasibility of the apical view compared with the subxiphoid view in TTE.
The proportion of mechanically ventilated patients in our population was 45%, which is lower than the proportion usually reported in ICU [
14]. However, in our study, TTE was performed within the first 2 days after admission of the patients and mostly at the time of admission. We do not believe that a higher proportion of mechanically ventilated patients would have modified our conclusions, as the proportion of adequate views was significantly higher for the apical view than for the subxiphoid view in the population of mechanically ventilated patients (Table
2).
The subxiphoid view is an essential element of echocardiography in an ICU, particularly to analyze the inferior vena cava (for example, size and collapsibility) and pericardium (for example, pericardial effusion). According to our results, only parameters recorded in two-dimensional mode (LVEF and RVEDA/LVEDA) can be recorded on the subxiphoid window without any bias. The bias of E/A was almost nil, but this ratio alone does not constitute a reliable parameter to analyze LV filling pressure in this setting of critically ill patients [
15]. The fairly large limits of agreement in the Bland-Altman representations for LVEF, E/A, and RVEDA/LVEDA can also be an argument in favor of the use of a semiquantitative approach for these parameters.
Figure
2 demonstrates the low rate of misclassified patients. Use of the subxiphoid window to analyze ventricular morphology therefore allows a point-of-care approach: preserved or altered LVEF, or normal or dilated right ventricle [
16,
17].
The discrepancy between apical and subxiphoid values for E, A, E/Ea, and Ao VTI can be explained by the angle between the Doppler axis and blood flow or myocardial movement in the various acoustic windows. With the Doppler mode, red blood cell velocity (or myocardial tissue movement on tissue Doppler) depends on the cosine of the angle between the axis of the Doppler beam and blood flow (myocardial tissue movement): the smaller the angle, the higher the velocity [
18]. On the apical view, the heart is examined in the longitudinal (anteroposterior) axis, ensuring perfect alignment between the Doppler beam and blood flow through the mitral or aortic rings and maximum velocities. On the subxiphoid view, the heart is viewed in a different axis (inferoanterior). Consequently, on this view, the alignment between the Doppler beam and aortic flow is not perfect, and the velocity reported by the echocardiographer will be underestimated. Interestingly, the E/A ratio is not modified on the subxiphoid view, as the E and A waves are both decreased to a similar degree. The velocity of the mitral annulus (Ea) is measured in tissue Doppler mode. On the apical view, the Doppler beam is aligned with the longitudinal movement of the mitral annulus, but the subxiphoid view measures the radial movement of the annulus, accounting for the lower Ea velocity obtained in this window. It is noteworthy that the E/Ea and Ao VTI were also underestimated in patients with a history of COPD and in mechanically ventilated patients, two situations theoretically associated with a vertical heart axis that could have resulted in better alignment on the subxiphoid view.
An adequate apical view could not be obtained in the present study in fewer than 15% of cases, whereas use of the subxiphoid view was incomplete, as Doppler analysis appears to be unreliable. Therefore, in the unusual situation of absence of an apical window, the use of another monitoring device should be considered (particularly in intubated patients, in whom transesophageal echocardiography can usually be performed).
This study presents several limitations. The lower velocities obtained on the subxiphoid view are due to the poor alignment between the Doppler axis and blood flow. Some echocardiography devices apply a correction of this angle by aligning the Doppler beam with blood flow, but this angle correction was not tested in this study. However, accurate determination of the appropriate angle correction for velocity is difficult, essentially because this angle can be measured only in a two-dimensional plane, which therefore disregards the angle in the third dimension of the heart. Moreover, angle correction would not be sufficient to correct the velocity of Ea, as the observed decrease is not exclusively due to poor alignment, but also because this view measures a different relaxation property of the myocardium: radial instead of longitudinal. Not all hemodynamic parameters were evaluated by TTE in this study. Right ventricular function was not extensively analyzed, apart from RVEDA/LVEDA (an important parameter of biventricular interaction). However, the misalignment observed in Doppler mode would also be expected to affect measurement of the tricuspid proto-systolic S-wave velocity on tissue Doppler, tricuspid annular plane systolic excursion in M mode (TAPSE), or pulmonary artery pressure. We also did not study the variation of aortic blood flow during passive leg raising or mechanical ventilation to predict fluid responsiveness, as this would have required a strictly selected population [
19]. Although the position of the heart and consequently our findings should not differ according to the reason for admission, it would be interesting to confirm our conclusions in a specific population of acute circulatory failure patients.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JM and AS contributed to the acquisition and analysis of data and drafting the manuscript
JM and MS contributed to the conception and design of the study. CT, ZM, GC, and MS were involved in interpretation of data, critically revising the manuscript, and giving final approval. All authors read and approved the final manuscript.