Introduction

Transferring a patient to the passive leg raising (PLR) position (in which the lower limbs are elevated at 45° while the trunk is lying supine) transfers venous blood from the legs to the intrathoracic compartment and increases cardiac preload [1]. The effects of this “endogenous fluid challenge” on aortic blood flow [2, 3] or cardiac output [4, 5] enable to test fluid responsiveness with accuracy. As a result, leg raising has been recommended as a part of hemodynamic monitoring by recent international recommendations [6].

In the previous studies dealing with PLR, the postural change ended at a similar PLR position (lower limbs elevated at 45° and trunk in supine position). However, at baseline, the position was either semi-recumbent (trunk elevated at 45° (PLRSEMIREC) [3, 4] or at 30° [5]) or supine (PLRSUPINE) [2, 7] (Fig. 1). Compared to PLRSEMIREC, PLRSUPINE could have a lower hemodynamic impact since it should not recruit the blood of the splanchnic reservoir. Nonetheless, PLRSUPINE could be considered as an alternative to PLRSEMIREC in some particular conditions such as increased abdominal pressure in which PLRSEMIREC can be contraindicated.

Fig. 1
figure 1

Study protocol with the two sequences of postural changes at which the patients were randomized. PLR Passive leg raising, PLR SEMIREC passive leg raising performed by elevating the patient’s legs and by simultaneously transferring the trunk from the semi-recumbent position to a horizontal position, PLR SUPINE passive leg raising performed by elevating the patient’s legs from the supine position

We compared the hemodynamic effects of PLRSEMIREC and of PLRSUPINE [8] and compared the ability of the two methods for predicting fluid responsiveness in view of standardizing the technique.

Materials and methods

Study population

We included patients if:

  1. 1.

    They presented an acute circulatory failure due to sepsis [9] (see ESM),

  2. 2.

    A prior PLRSEMIREC maneuver had increased the pulse contour-derived cardiac output (PiCCOplus® device v6.0, Pulsion Medical Systems™, Munich, Germany) by more than 10%.

Non-inclusion criteria were deep venous thrombosis or elastic compression stocking, head trauma or an increase in the intra-abdominal pressure suspected by clinical context and examination. We included 35 patients whose characteristics are listed in Table 1. All patients or next of kin gave their deferred consent.

Table 1 Characteristics of the patients (n = 35)

Study design

After the initial PLRSEMIREC, the patients were transferred in the semi-recumbent position. They were then submitted to the following postural maneuvers (Fig. 1):

  • a PLRSEMIREC which consisted of transferring the patients from the 45° semi-recumbent position to the PLR position;

  • a “supine transfer” which consisted of transferring the patients from the 45° semi-recumbent position to the supine position followed by a PLRSUPINE which consisted of transferring the patients from the supine position to the PLR position (Fig. 1).

These two sequences of postural changes were 2-min long each [3, 4] and were performed successively in each patient in a randomised order. Finally, 500 mL saline were administered intravenously over 10 min.

Measurements

After each position change and after volume expansion, we recorded the heart rate, the systemic arterial pressure, the pulse contour-derived cardiac output (PiCCOplus® device v6.0, Pulsion Medical Systems™, Munich, Germany), the central venous pressure and the right ventricular end-diastolic area by an apical 4-chamber view at transthoracic echocardiography (EnVisor B.0, Philips Medical System, Andover, MA). The pulse contour-derived cardiac output was calibrated by transpulmonary thermodilution at baseline.

Data analysis

Patients in whom the fluid administration induced an increased cardiac index by more than 15% were defined as “responders” to fluid administration [3, 1012]. In an attempt to estimate the endogenous volume recruited by the PLRSEMIREC, we noted during fluid infusion the volume that had been infused at the time when cardiac output precisely reached the value that it had reached during the PLRSEMIREC maneuver. We tested whether fluid responders were detected by a PLRSUPINE-induced increase in cardiac index by more than 10%, which was the cut-off value of cardiac index increase in response to PLRSEMIREC previously reported to discriminate fluid responders and non responders [3, 4].

Comparisons of hemodynamic variables between before versus during PLR, between before versus after supine transfer and between before versus after volume expansion, comparisons of the changes in hemodynamic variables induced by PLRSUPINE, PLRSEMIREC, supine transfer and volume expansion were assessed using a paired Student t test (Statview 5.0 software, Abacus concepts, Berkeley, CA). Data were expressed as median (25–75% interquartile range). A P value ≤ 0.05 was considered statistically significant.

Results

Results are expressed like if in all patients the sequence of postural changes would have been: PLRSEMIREC, supine transfer and then PLRSUPINE.

The PLRSEMIREC maneuver, the supine transfer and the PLRSUPINE maneuver increased cardiac index by 22 (17–28), 9 (5–15) and 10 (7–14)%, respectively (Table 2; Fig. 2).

Table 2 Evolution of hemodynamic variables during the different postural changes
Fig. 2
figure 2

Changes in cardiac index (CI) induced by the different postural changes, expressed in % change [median (25–75IR)] from the value measured before the postural change, n = 35; PLR Passive leg raising, PLR SEMIREC passive leg raising performed by elevating the patient’s legs and by simultaneously transferring the trunk from the semi-recumbent position to a horizontal position, PLR SUPINE passive leg raising performed by elevating the patient’s legs from the supine position

The PLRSEMIREC maneuver increased the right ventricular end-diastolic area by 20 (14–29)% and the central venous pressure by 33 (22–50)%. The supine transfer increased the right ventricular end-diastolic area by 9 (5–16)% and the central venous pressure by 15 (10–20)%. The PLRSUPINE maneuver increased the right ventricular end-diastolic area by 10 (5–16)% and the central venous pressure by 20 (15–29)% (Table 2; Fig. 3).

Fig. 3
figure 3

Changes in central venous pressure (CVP) induced by the different postural changes, expressed in % change [median (25–75IR)] from the value measured before the postural change, n = 35; PLR Passive leg raising, PLR SEMIREC passive leg raising performed by elevating the patient’s legs and by simultaneously transferring the trunk from the semi-recumbent position to a horizontal position, PLR SUPINE passive leg raising performed by elevating the patient’s legs from the supine position

Volume expansion significantly increased cardiac index by 27 (21–38)%. All patients were responders to volume expansion. In all patients, the PLRSEMIREC maneuver of the protocol had increased cardiac index by more than 10%, i.e. that maneuver reproduced the effects of the PLRSEMIREC performed before inclusion in the study. The effects of the PLRSEMIREC on cardiac index were reached when 312 (250–350)mL of saline had been infused.

The PLRSUPINE increased cardiac index by more than 10% in only 20 patients. Thus, if an increase in cardiac index by more than 10% would have been considered as a positive response to PLRSUPINE as it is used for PLRSEMIREC [3], 15 (43%) patients would have been unduly predicted as non-responders to fluid administration by a PLRSUPINE test. In these 15 patients, the PLRSUPINE had increased cardiac index by 5 (2–8)% while the subsequent volume expansion increased cardiac index in a lower extent than in the other ones [23 (19–27) vs. 33 (24–45)%, respectively, P < 0.05].

Discussion

This study demonstrated (1) that cardiac preload and cardiac index increased more during PLRSEMIREC than during PLRSUPINE, (2) that the 10% increase in cardiac index, which has been previously found as the diagnostic cut-off for predicting fluid responsiveness by the PLRSEMIREC test [3, 4], could not be used for the PLRSUPINE test.

Lifting the legs up to 45° was supposed to transfer venous blood from the legs toward the intrathoracic compartment [13], increasing the intrathoracic blood volume [14] and inducing an increase in the cardiac preload [7, 15] as confirmed in the present study by the marked increase in central venous pressure and in the right ventricular end-diastolic area during PLR. As a clinical application, PLR has been developed as an “endogenous fluid challenge” [16, 17] that we found to be of around 300 mL. PLR predicts fluid responsiveness with accuracy, especially in case of arrhythmias or spontaneous breathing [35].

However, there is a debate regarding the baseline posture before moving the patient to the final PLR position (lower extremities at 45° and trunk supine) [1]. The present study evidenced that PLRSEMIREC exerted larger “auto-fluid loading” effect than PLRSUPINE. All patients of the study were preload responsive and, as a consequence of a larger increase in preload, the increase in cardiac index was significantly larger during PLRSEMIREC than during PLRSUPINE. Interestingly, the difference in cardiac preload and cardiac index changes found between the two postural maneuvers corresponded to those induced by transferring the patient’s trunk from the semi-recumbent to the supine position before performing the PLRSUPINE, suggesting that PLRSEMIREC induced the additional recruitment of the vast splanchnic reservoir.

Importantly, if a 10% increase in cardiac index was used for defining a “positive” PLRSUPINE test, it would have falsely classified more than 40% of the patients as fluid non-responders while they actually were responders. In those “false negative” patients, volume expansion increased cardiac output by a lower extent than in the other ones, indicating that their preload dependence was of lower degree and suggesting that the PLRSUPINE test was not sensitive enough to detect it.

The main limitation of our study is that for ethical reasons, we included only responders to volume expansion such that it was not be possible to directly compare the accuracy of the PLR methods and to determine the cut-off for the PLRSUPINE test. We did not recalibrate pulse contour cardiac output by thermodilution at each step of the protocol to avoid the injection of a large amount of fluid. Nevertheless, the potential resulting error in cardiac output measurement likely affected both postural methods, such that it should have not altered our results. Moreover, the pulse contour analysis, which was reported to be far more reproducible than transpulmonary thermodilution [18] should be quite appropriate to allow a correct interpretation of the PLR tests [19]. Finally, we administered the same amount of fluid in all patients, what should have been excessive in some patients and not enough in some others. In this regard, it should be advised to repeat PLR at the end of fluid administration for testing the need for additional fluid administration.

To summarize, when PLR started from the supine position, its increasing effects on cardiac preload and cardiac index were lower than if it started from the semi-recumbent position. Moreover, the same cut-off value could not be used interchangeably for the two PLR methods. This strongly suggests that the PLRSEMIREC should be preferred for assessing fluid responsiveness in patients with septic circulatory failure.