Passive leg raising test using the carotid flow velocity–time integral to predict fluid responsiveness

Abstract

Purpose

The passive leg raising test (PLR) is a noninvasive method widely adopted to assess fluid responsiveness. We propose to explore if changes in the carotid flow assessed by echo-Doppler can predict fluid responsiveness after a PLR.

Methods

We conducted a performance diagnostic study in two intensive care units from Argentina between February and April 2022. We included patients with signs of tissular hypoperfusion that required fluid resuscitation. We labeled the patients as fluid responders when we measured, after a fluid bolus, an increase greater than 15% in the left ventricle outflow tract (LVOT) VTI in an apical 5-chamber view and we compared those results with the carotid flow (CF) velocity–time integral (VTI) from the left supraclavicular region in a semi-recumbent position and during the PLR.

Results

Of the 62 eligible patients, 50 patients (80.6%) were included. The area under the ROC curve for a change in CF VTI during the PLR test was 0.869 (95% CI 0.743–0.947). An increase of at least of 11% in the CF VTI with the PLR predicted fluid-responsiveness with a sensitivity of 77.3% (95% CI 54.6–92.2%) and specificity of 78.6% (95% CI 59–91.7%). The positive predictive value was 73.9% (95% CI 57.4–85.6%) and the negative predictive value was 81.5% (95% CI 66.5–90.7%). The positive likelihood ratio was 3.61 and the negative likelihood ratio was 0.29.

Conclusion

An increase greater than 11% in CF VTI after a PLR may be useful to predict fluid responsiveness among critically ill patients.

Introduction

The main aim of fluid resuscitation is to increase the cardiac output and enhance organ perfusion, but since positive fluid balance is associated with increased mortality it must be guided by a reliable parameter, especially considering only 50% of patients are fluid responders [1–4].

Static parameters, such as central venous pressure, have not been shown to be accurate predictors of fluid responsiveness, as they have poor correlation with intravascular volume and cardiac function [5].

Passive leg raising test (PLR) has gained popularity as a noninvasive and reversible method to assess fluid responsiveness. This is because it produces a mobilization of 200 to 300 mL of blood from the venous reservoir of the lower extremities to the thorax, thereby increasing venous return and challenging the ventricular Frank-starling curve. The purpose of the PLR is to determine whether a patient is preload dependent, and therefore likely to benefit from fluid resuscitation. This test has been validated in numerous studies [6].

As the preload augmentation induced by PLR is reversible and ephemeral, the hemodynamic response should be evaluated within 30 to 90 s after the maneuver. This is because the blood that was displaced from the legs will eventually return to the lower extremities, reducing the venous return and cardiac output to baseline levels. Therefore, the measurement of cardiac output should be performed promptly and in real-time [6, 7].

Transthoracic echocardiography (TTE) is an optimal tool for this assessment, it enables to estimate the stroke volume (SV) by measuring the velocity–time integral (VTI) of the left ventricular outflow tract (LVOT) in an apical 5-chamber view. Nevertheless, critically ill patients often pose a challenge to obtain an apical 5-chamber view due to the patient position and the mechanically ventilated air-filled lungs [8–10].

Several studies have proposed as an alternative the measurement of the flow of the common carotid artery as a surrogate of the SV using a conventional method with the linear transducer [11–14]. This technique has some caveats. The Doppler’s signal angle of incidence from the carotid flow direction is not in the optimal range of 0°. The angle correction function should be utilized in this situation, accepting a maximum angle of 60° [15]. Considering these limitations, we have previously described a novel approach to measure the carotid systo-diastolic flow VTI by placing a phase array transducer at the level of the left supraclavicular fossa [16].

We proposed to evaluate if the carotid flow (CF) measured by echo-Doppler, adopting the previously described approach that improves the technical limitations, can be used to detect the hemodynamic changes produced by the PLR in order to predict fluid responsiveness.

Materials and methods

This is a diagnostic accuracy study, carried out in Sanatorio de Los Arcos, a 20 medical-surgical bed Intensive Care Unit (ICU), and Hospital Alemán, a 30 medical-surgical bed ICU in Buenos Aires, Argentina, from February to April 2022. The Institutional Review Board (Comité de Ética en Investigación, Clínica y Maternidad Suizo-Argentina) approved the study and written informed consent was obtained from the patients or their representatives.

We included patients older than 18 years old admitted to the Intensive Care Unit who, according to the physician's discretion, would benefit from fluid bolus due to signs of tissue hypoperfusion. These included: arterial hypotension (mean arterial pressure less than 65mmHg), tachycardia (heart-rate greater than 100 beats per minute), oligoanuria (diuretic rhythm less than 20 ml/hr), and increased lactate (greater than 20mg/dl). Exclusion criteria were inadequate ultrasonographic image quality, inability to align Doppler signal with LVOT in an apical 5-chamber view and/or carotid flow at left supraclavicular region, LVOT dynamic obstruction, significant aortic valve disease (moderate and/or severe stenosis and/or regurgitation), aortic dissection, cardiac arrhythmia, abdominal hypertension, lower limb fracture, intracranial hypertension, significant carotid disease (stenosis greater than 50%), and cerebral circulatory arrest.

Measurements

We obtained recordings of heart rate, invasive or noninvasive blood pressure, oxygen saturation and ECG signal from multiparametric monitorization.

For ultrasonographic evaluation we used a Philips Sparq ultrasound machine (Philips Healthcare, Bothell, WA, USA) with a phased array transducer with a frequency range of 2 to 4 MHz, and a Philips CX50 (Philips Healthcare, Bothell, WA, USA) with a phased array transducer with a frequency range of 2 to 4 MHZ.

Carotid flow assessment was performed by a physician certified in ultrasonography and critical echocardiography (IC, FS). Carotid flow measurements were obtained by placing the transducer at the level of the left supraclavicular region, pointing the probe towards podalic in order to evaluate the carotid flow near its origin in the aorta. With the aid of color Doppler, the left common carotid artery was identified as a tubular structure with its flow approaching the probe (Fig. 1). The pulsed wave Doppler was activated, and the sample volume was located as proximal as possible to the carotid’s origin, in the aortic arch. With this view, it was possible to obtain a perfect alignment between the direction of flow and the incidence of the Doppler signal. The spectra of three consecutive VTI of both systolic and diastolic flow were recorded (Fig. 2). We did not used the Doppler angle correction function in any measurement.

Fig. 1.

Position of the phased array transducer at the level of the left supraclavicular region to measure the carotid flow

Fig. 2.

A Color Doppler mode of the left common carotid artery that is perfectly aligned with the dotted line representing the incidence of the pulsed Doppler signal. B Measurement of 3 CF VTI consecutively

Echocardiographic evaluation was performed by a physician certified in ultrasonography and critical echocardiography at each center (VOC, BOT). In an apical 4-chamber view, LV systolic function was assessed qualitatively and quantitatively with the biplane Simpson method. Color and continuous Doppler was used at the valve level to determine the presence of significant valvulopathies. Right ventricular size and systolic function were also assessed visually and with a TAPSE measurement. In the apical 5-chamber view, LVOT VTI was measured with the pulsed wave Doppler mode just above the aortic valve, while striving to maintain the angle of the Doppler signal to LVOT as close to 0°, ensuring that a linear signal also appeared immediately after the flow corresponding to the aortic valve closure click, obtaining 3 consecutive values for estimate an average.

Study protocol

We recorded the systolic arterial pressure (SAP), diastolic arterial pressure (DAP), mean arterial pressure (MAP), heart rate (HR), and LVOT VTI and CF VTI under baseline conditions with the patient’s head of the bed at 45° (semi-recumbent position). Later, the patient was positioned at 0° (laid flat) and both lower limbs were elevated to 45° with the aid of customed cushions. After 1 min, a second set of measurements of SAP, DAP, MAP, HR, and CF VTI were recorded (Fig. 3). The patient was once again placed in a semi-recumbent position, and infused with 500 mL of Ringer Lactate over 15 min. A third set of measurements (SAP, DAP, MAP, HR, and LVOT VTI) were recorded after the fluid bolus. The patients that presented an increase of at least 15% in the LVOT VTI after the fluid bolus were classified as responders [17]. Ventilation settings and the vasoactive drugs were not modified during the study protocol.

Fig. 3.

CF VTI measurement with the patient in semi-recumbent position (A), and during the PLR (B)

Recording of data

The following data were collected from the electronic health record: age, gender, reason for hospitalization, APACHE II score on admission, requirement of invasive mechanical ventilation, respiratory rate, oxygen saturation by pulse oximeter, HR, blood pressure. The LV ejection fraction by biplane Simpson method, TAPSE, LVOT VTI, CF VTI were recorded in the ultrasonography machine and subsequently loaded by the investigators appointed at each center into specific forms.

Statistical analysis

The MedCalc^® Statistical Software version 20.011 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2021) with the sample size function for area under the receiver operating characteristic (ROC) curve was used to calculate the sample size. Assuming a power of 0.8 and an alpha value of 0.05, for the parameters area under the ROC curve = 0.75, null hypothesis value = 0.5, ratio of sample sizes in negative/positive groups = 1, the calculated sample size was 38.

The frequencies and distribution of qualitative variables were expressed, according to the observed characteristics of the distribution, as absolute numbers and percentages, while quantitative variables were expressed as means ± standard deviations or medians and interquartile ranges. The normal distribution of quantitative variables was evaluated visually with frequency histograms and analytically with the Shapiro-Wilks test. Associations between numerical variables were studied with a Student’s t test or Mann–Whitney test. Associations between qualitative variables were studied using the Chi-square test or Fisher’s exact test, depending of the expected frequency of observations. Paired t test and Wilcoxon rank sum test were used to compare pre- and post-PLR data, as appropriate.

An area under the ROC curve was constructed using the Hanley-McNeil test to determine the cutoff for the best sensitivity and specificity for the percentage of change in CF VTI with PLR in identifying the fluid responder patients (those who exhibited an increase in LVOT VTI of at least 15% after fluid expansion).

For all measurements, a 95% CI difference showing no overlap or a p-value less than 0.05 was considered statistically significant.

Results

Of the 62 eligible patients, 50 patients (80.6%) were included between February and April 2022. Six patients were excluded due to inadequate cardiac ultrasonographic image quality, three were excluded due to moderate aortic valve regurgitation, two were excluded because of an inability to align Doppler signal with LVOT in an apical 5-chamber view, and one was excluded because of LVOT dynamic obstruction. The median age was 68 years, 64% of the patients were male. The most frequent causes of hospitalization were surgery (32%), sepsis (16%), septic shock (12%), and pneumonia (6%). The median value of the APACHE II score was 12, 46% of the patients were under mechanical ventilation, 52% required Noradrenaline and 2% Dobutamine. Regarding the echocardiographic parameters, 66% of the patients had a preserved left ventricular systolic function, and most of the patients (90%) had a preserved right ventricular systolic function. Among the 50 patients, 22 (44%) were fluid responders and 28 (56%) were non-responders (Table 1).

Table 1.

Demographic and clinical characteristics of the patients (n = 50)

Variable	Global N = 50	Responders N = 22	Non-responders N = 28	p value
Age (years)1	68 (58–74)	64 (50–73)	69 (62–76)	0.225
Male, n (%)	32 (64)	12 (54.5)	19 (67.8)	0.341
Diagnosis, n (%)
Surgery	16 (32)	8 (36.3)	8 (28.6)	0.566
Sepsis	8 (16)	5 (22.7)	3 (10.7)	0.255
Septic shock	6 (12)	3 (13.6)	3 (10.7)	0.756
Pneumonia	3 (6)	2 (9.1)	1 (3.6)	0.421
Pancreatitis	2 (4)	1 (4.5)	1 (3.6)	0.873
Intracerebral hemorrhage	2 (4)	1 (4.5)	1 (3.6)	0.873
Other	13 (26)	2 (9.1)	11 (39.3)	0.016
APACHE II score1	12 (8–18)	10.5 (8–16)	14 (9–16)	0.43
Mechanical ventilation, n (%)	23 (46)	11 (50)	12 (42.8)	0.615
Vasopressors, n (%)	26 (52)	14 (63.6)	12 (45.8)	0.214
Inotropic, n (%)	1 (2)	1 (4.5)	0	0.261
LV systolic function, n (%)
Normal	33 (66)	13 (59.1)	20 (71.4)	0.367
Mildly impaired	14 (28)	7 (31.8)	7 (25)	0.277
Moderately impaired	2 (4)	2 (9.1)	0	0.106
Severely impaired	1 (2)	0	1 (3.6)	0.373
RV systolic function, n (%)
Normal	45 (90)	18 (81.8)	27 (96.4)	0.091
Impaired	5 (10)	4 (18.2)	1 (3.6)	0.089

¹Median (IQR)

Fluid responders presented a statistically significant increase in systolic, diastolic and mean arterial pressure during PLR. A significant increase in the CF VTI value during PLR was also noted in both fluid responders and non-responders (Table 2).

Table 2.

Hemodynamic parameters at baseline and after PLR in responders and non-responders

	Baseline	During PLR	p value
HR (beats/min)1
Responders	101 (93–107)	97.5 (89–103)	0.104
Non-responders	86 (76–97.5)	88 (78–98.5)	0.418
SAP (mmHg)1
Responders	106 (96–112)	119 (105–129)	0.001
Non-responders	122 (100–139.5)	121.5 (107.5–136)	0.25
DAP (mmHg)1
Responders	58 (52–62)	66 (59–70)	0.002
Non-responders	64.5 (56–71.5)	65.5 (60–71)	0.21
MAP (mmHg)1
Responders	72 (65–77)	81.5 (74–86)	0.001
Non-responders	76 (66–87)	79 (71–86)	0.414
CF VTI
Responders	16.3 (15.7–23)	19.7 (18.9–25.7)	< 0.0001
Non-responders	19.2 (16.9–24.7)	21.8 (18.5–25.3)	< 0.0001

¹Median (IQR). HR heart rate. SAP systolic arterial pressure. DAP diastolic arterial pressure. MAP mean arterial pressure. CF VTI carotid flow velocity–time integral. PLR passive leg raising test

Among responders, the percentual change of the CF VTI with the PLR was 18.1% (IQR 12.2–26.2), while in non-responders was 5% (IQR 2.6–10) with statistically significant difference (p < 0.0001) (Fig. 4).

Fig. 4.

Boxplot and individual values of the percentual change of the CF VTI during the PLR test among responders (R) and non-responders (NR)

The area under the ROC curve for the percentual change in CF VTI induced by the PLR was 0.869 (95% CI 0.743–0.947). An increase greater than 11% of the CF VTI after the PLR predicted fluid responsiveness with a sensitivity of 77.3% (95% CI 54.6–92.2%) and specificity of 78.6% (95% CI 59–91.7%). The positive predictive value (PPV) was 73.9% (95% CI 57.4–85.6%) and negative predictive value was 81.5% (95% CI 66.5–90.7%). The positive likelihood ratio was 3.61 (95% CI 1.71–7.59) and the negative likelihood ratio was 0.29 (95% CI 0.13–0.647) (Fig. 5).

Fig. 5.

Area under the ROC curve for the change in CF VTI with PLR

Discussion

Identifying which patients will benefit from fluid resuscitation is crucial, as administering fluids to non-fluid responders may lead to ineffective positive fluid balance and increase mortality [1–3]. The PLR can simulate the hemodynamic response of a 300 ml fluid bolus by producing a reversible shift of blood from the lower limb and pelvis to the intrathoracic compartment, without the risk of volume overload due to its reversible effect [3, 6, 7].

One of the preferred methods to determining fluid responsiveness is to perform a PLR with real-time stroke volume measurement using TTE [17]. Most studies that have assessed fluid responsiveness with PLR by TTE have relied on changes in LVOT VTI or SV obtained in an apical 5-chamber view. However, achieving an adequate apical 5-chamber view can be a challenge in critically ill patient, which remains the main obstacle to measuring LVOT VTI in this population [8, 9].

The carotid flow has been the object of investigation as a potential surrogate of cardiac output because is easier to asses than echocardiography. However, it is more prone to obtain erroneous measurements due incorrect alignment of the insonation angle [18, 19]. In the LVOT VTI measurement, usually the Doppler incidence aligns perfectly with the direction of the blood flow. In contrast, in carotid Doppler, the angle of incidence is wider due to the placement of the artery. In most cases, it is around 60° when using the high-frequency linear transducer, and achieving a 0° angle with a linear probe is generally not possible. When performing a Doppler exam, the measurements are reliable when the insonation is in a 0° angle from the flow, and not accurate when it approaches 90°. For this reason, the steering angle function is used with the linear transducer to attempt to reduce the angle and estimate the correct velocity. However, its function is still limited and can lead to incorrect values [5].

The use of CF VTI to predict fluid responsiveness has shown contradictory results. Girotto et al. reported that the carotid VTI variation after PLR maneuver had an area under the ROC curve of 0.58 to predict a significant increase in cardiac index, suggesting that it may not be reliable to assess cardiac output and its changes [3].

Marik et al. studied the use of carotid Doppler to determine fluid responsiveness using the PLR. They reported that the increase of carotid flow in responders had an excellent correlation with the SV variation with the PLR. However, it should be noted that they used thoracic bioreactance to measure the SV index as the gold standard. Its accuracy is questioned in different studies [10, 20].

Some authors have surpassed the problem of the angle of insonation by using the carotid artery flow time (CFT), a parameter that is angle independent. The CFT measured by pulsed wave Doppler mode is the time in milliseconds from the onset of systolic flow to the dicrotic notch. This value represents the duration of the left ventricular systolic ejection phase, which depends on preload, afterload and inotropism. Therefore, CFT needs to be corrected by heart rate (CFTc) [1, 22]. Unfortunately, the studies using CFTc as a surrogate of SV to assess fluid-responsiveness also have shown dissimilar results [5]. Karadadaş et al. obtained a cut-off value of 28 ms of CFTc variation to predict fluid responsiveness in patients with gastrointestinal bleeding with a sensitivity of 88% and specificity of 69% [2]. Jalil et al. described that an increase in CFTc of 24.6% with PLR predicted fluid responsiveness in critically ill patients, with a sensitivity of 60% and specificity of 92% [21]. However, Ma et al. reported that the correlation of CFTc with the cardiac output measured by thermodilution was poor, as well as the detection of fluid responsiveness [11]. Kenny et al. investigated the ability of CF VTI and CFTc, measured using a wearable Doppler ultrasound, to predict a 10% increase in stroke volume. The findings revealed an area under the ROC curve of 0.97 and 0.98 respectively. However, it is essential to consider the study’s limitations, such as the small sample size of only 11 healthy volunteers, and using non-invasive pulse contour analysis as a gold standard [22].

In consideration of the previously mentioned difficulties with the carotid Doppler insonation, we recently reported the use of a novel technique to assess the carotid flow at the level of the left primitive carotid artery using a phase-array transducer instead of the linear one [16]. This approach allowed us to obtain an optimal insonation angle between the Doppler signal and the blood flow direction.

The present study suggests that changes in CF VTI with PLR could predict fluid responsiveness in critically ill patients. A cut-off value of 11% to identify fluid responders has a sensitivity of 77.3% and specificity of 78.6%. To our knowledge, this is the first study to assess variations in carotid flow after PLR with this novel approach.

Although CF VTI should not be considered a substitute for echocardiography, it does provide a valuable alternative method that can complement the assessment of fluid responsiveness in critically ill patients. This can be especially crucial in patients with limited access to echocardiography or other methods of measuring cardiac output, or in situations where echocardiography is not feasible.

Our study presents several limitations. It was carried out just in two centers, which could threaten its generalizability. We did not take into account the inter- and intra-observer variability. We only measured the flow of the left primitive carotid artery, assuming that it should reflect more accurately the LV SV as it arises directly from the aorta, as opposed to the right side, which arises from the innominate artery. Nevertheless, further studies are needed to draw conclusions. We did not measure the CFTc to examine the potential benefits of using a combination of a CF VTI and CFTc to enhance sensitivity and specificity. It would have been interesting to observe the results of such an investigation, and this possibility could be explored in future research. Lastly, we categorize as fluid-responsive the patients with a variation greater of 15% in LVOT VTI after a fluid bolus infusion, this cut-off value was chosen from prior studies and was not confirmed in this study [5, 17–19].

Conclusion

An increase greater than 11% in CF VTI after a PLR may be useful to predict fluid responsiveness among critically ill patients.

Author contributions

Conceptualization: IC; Methodology: IC; Formal analysis: IC; Investigation: IC, VOC, FAS, BTO, MFF; Writing—original draft preparation: IC; Writing—review and editing: IC, VOC; Supervision: IC, FMT, PMM.

Data availability

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Declarations

Conflict of interest

The authors have no relevant financial or non/financial interests to disclose.

Ethical approval

The study was conducted following the principles of the Helsinki Declaration and within the precautions established by ethical and legal standards. The study was approved by the Institutional Review Board and Ethic Committee of Clínica y Maternidad Suizo-Argentina (approval number 6598).

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Consent to publish

The authors affirm that human research participants provided informed consent for publication of the images in Figs. 1, 2, 3, and 4