Trendelenburg position is a reasonable alternative to passive leg raising for predicting volume responsiveness in mechanically ventilated patients in the ICU

Abstract

Background

Passive leg raising (PLR) is a well-recognized method for assessing volume responsiveness in the intensive care unit (ICU); however, it has some limitations. The physiology of the cardiac preload induced by the Trendelenburg position and PLR is similar. The Trendelenburg position can be initiated from the supine position and then tilted downward (TREND_SUPINE) or from the reverse Trendelenburg position and then tilted further downward (TREND_rTREND). Therefore, this study aimed to compare the predictive performance of the percentage change in stroke volume index (ΔSVI) induced by TREND_SUPINE, TREND_rTREND, and PLR for volume responsiveness in mechanically ventilated patients in the ICU.

Methods

The study was a prospective single-center cohort study conducted in a comprehensive ICU. The study consisted of the following sequential steps: (1) baseline-1: supine position with a 0° bed angulation; (2) Trendelenburg position: 15° downward bed angulation; (3) baseline-2: the same position as baseline-1; (4) reverse Trendelenburg position: 15° upward bed angulation; (5) Trendelenburg position: 15° downward bed angulation; (6) baseline-3: the same position as baseline-1; (7) semi-recumbent position: trunk elevated at 45°; (8) PLR: lower limbs elevated at 45° and trunk in the supine position; (9) baseline-4: the same position as baseline-1; (10) volume loading test: 250 ml of 4% albumin was infused over 10 min. At each time point, hemodynamic parameters were monitored using pulse contour analysis. A positive volume response was defined as an increase of at least 15% in SVI on pulse contour monitoring from baseline-4 after the volume loading test. The receiver operating characteristic curves were performed for ΔSVI.

Results

In this study, 36 patients were included for analysis, and of these, 15 patients (42%) were volume-responsive. TREND_SUPINE-induced ΔSVI (ΔSVI_SUPINE-TREND), TREND_rTREND-induced ΔSVI (ΔSVI_rTREND-TREND) and PLR-induced ΔSVI (ΔSVI_PLR) were 6%, 16%, and 11%, respectively (ΔSVI_SUPINE-TREND vs. ΔSVI_{rSUPINE-TREND}, p < 0.001; ΔSVI_SUPINE-TREND vs. ΔSVI_PLR, p < 0.05). The R² values of the linear regression between ΔSVI_SUPINE-TREND, ΔSVI_rTREND-TREND, ΔSVI_PLR and volume loading test-induced ΔSVI were 0.14, 0.32, and 0.43, respectively (all p < 0.001). Furthermore, the area under the receiver operating characteristic curve for predicting volume responsiveness was 0.78 [95% confidence interval (CI), 0.59–0.92], 0.88 (95%CI, 0.75–0.96), and 0.83 (95%CI, 0.61–0.95) for TREND_SUPINE, TREND_rTREND, and PLR-induced ΔSVI, respectively, with no statistically significant difference among them. The sensitivity and specificity for predicting volume responsiveness were 93% and 67% for ΔSVI_SUPINE-TREND at 4% cutoff, 87% and 76% for ΔSVI_rTREND-TREND at 13% cutoff, and 73% and 86% for ΔSVI_PLR at 11% cutoff.

Conclusions

The Trendelenburg position (TREND_SUPINE and TREND_rTREND)-induced and PLR-induced percentage changes in SVI were similar in their ability to predict volume responsiveness in mechanically ventilated patients in the ICU. Considering that TREND_rTREND induced greater percentage changes in SVI, it is preferentially recommended as a reasonable alternative to PLR for predicting volume responsiveness in certain clinical scenarios.

Trial registration: ChiCTR2300067694. Registered on January 18, 2023.

Background

In the intensive care unit (ICU), accurate assessment of volume responsiveness and administration of appropriate fluids to patients is the cornerstone of hemodynamic management [1]. Meta-analyses have shown that fluid intervention strategies oriented to volume responsiveness assessment significantly reduced patient mortality and ICU length of stay [2]. Various methods have been used to assess volume responsiveness, including fluid challenge [3], end-expiratory occlusion test [4], and passive leg raising (PLR) test [5]. Among them, PLR, with the advantages of simplicity of operation, repeatability and reversibility, no additional fluids, and independence from spontaneous respiration and cardiac arrhythmia, is a commonly used method for assessing volume responsiveness in the ICU. Clinical studies have shown that PLR accurately predicts volume responsiveness [6–9]. However, PLR cannot be used in special circumstances, such as during surgery, hip dislocation, or after certain surgical procedures (e.g. in patients undergoing lower extremity surgery), or in patients ventilated in the prone position. If the patient is an amputee or has severe myasthenia or if the patient wears compression stockings, the increase in venous return by PLR is limited and insufficient to induce a significant increase in cardiac output (CO) [10, 11].

Similar to the principles of PLR, the Trendelenburg position is an autologous volume-loading test in which gravity increases blood return from the lower extremities and viscera to the heart [12]. The Trendelenburg position has been found to predict volume responsiveness in patients with acute respiratory distress syndrome (ARDS) in the prone position [13] as well as during veno-arterial extracorporeal membrane oxygenation (V-A ECMO) support [14]. However, no studies have compared the predictive effects of Trendelenburg position and PLR on volume responsiveness.

The Trendelenburg position can be initiated either from the supine position with a subsequent downward tilt of the body (TREND_SUPINE) or from the reverse Trendelenburg position followed by a further downward tilt (TREND_rTREND). Therefore, the main objective of this study was to compare the predictive performance of the percentage change in stroke volume index (ΔSVI) induced by TREND_SUPINE, TREND_rTREND, and PLR for volume responsiveness in mechanically ventilated patients in the ICU.

Methods

Study design and ethical approval

The study was conducted from January 20, 2023 to April 20, 2023 in the Department of Critical Care Medicine (30 beds) of Tianjin Third Central Hospital (Tianjin, China), and was registered with the China Clinical Trial Registry (registration number: ChiCTR2300067694). The study protocol was approved by the Ethics Committee of Tianjin Third Central Hospital (Ethics Approval Number: IRB2019-037-01). All proxies for the included patients provided informed consent.

Study population

The inclusion criteria were as follows: age ≥ 18 years; ongoing hemodynamic monitoring with the pulse indicator continuous cardiac output (PiCCO) device; presence of at least one of the following criteria for inadequate tissue perfusion [15]: systolic blood pressure < 90 mm Hg (or a drop in blood pressure of > 50 mm Hg in patients with hypertension), the need for vasoactive medication (dopamine > 5 μg.kg⁻¹.min⁻¹ or the need for norepinephrine application), arterial blood lactate of > 2 mmol.L⁻¹, urine output < 0.5 ml.kg⁻¹.h⁻¹ lasting at least 2 h, tachycardia (heart rate > 100 min⁻¹), and prolonged capillary refill time or skin floridities.

The exclusion criteria were as follows: high-risk of volume loading test (VLT) (e.g. acute coronary syndrome, heart failure, cardiogenic shock, or evidence of volume overload), pregnancy, cranial hypertension or risk of cranial hypertension (e.g. cerebrovascular accident, craniocerebral trauma), cardiac arrhythmia, and wearing compression elastic stockings.

Study protocol

All study participants received a combination of midazolam and remifentanil for analgesia and sedation, with a target sedation depth of a RASS (Richmond Agitation-Sedation Scale) of −5. Throughout the study period, the doses of analgesic and sedative drugs, vasoactive drugs, and ventilator parameters were maintained constant. The study consisted of ten consecutive steps: (1) baseline-1: supine position with a 0° bed angulation; (2) Trendelenburg position: 15° downward bed angulation [16]; (3) baseline-2: the same position as baseline-1; (4) reverse Trendelenburg position: 15° upward bed angulation; (5) Trendelenburg position: 15° downward bed angulation; (6) baseline-3: the same position as baseline-1; (7) semi-recumbent position: trunk elevated at 45°; (8) PLR: lower limbs elevated at 45° and the trunk in the supine position; (9) baseline-4: the same position as baseline-1; (10) VLT: 250 ml of 4% albumin was infused over 10 min. Parameters recorded at each step included heart rate (HR), blood pressure (BP), stroke volume index (SVI), continuous cardiac index (CCI), pulse pressure variation (PPV), stroke volume variation (SVV), and central venous pressure (CVP). Each step from steps (1) to (9) lasted 1 min, and parameters were recorded 1 min after stabilization in steps (1), (3), (6), and (9) and immediately after VLT in step (10). In steps (2), (5), and (8), parameters were recorded at the time of maximum SVI within 1 min, and in steps (4) and (7), parameters were recorded at the minimum SVI within 1 min. The study protocol is illustrated in Fig. 1.

Fig. 1.

Study program. BA: bed angulation, rTrendelenburg: reverse Trendelenburg, PLR: passive leg raising

The following adverse events were closely monitored during the study: decrease in arterial systolic BP > 30 mmHg, increase in HR > 10%, pulse oximetry < 88%, new-onset arrhythmia, or any adverse event deemed by the investigator to be study-related [13]. If any of these adverse events occurred, the study was terminated immediately.

Measurements

All patients had a central venous catheter in the superior vena cava and a thermistor-tipped arterial catheter inserted via the femoral artery. A PiCCO₂ device (Pulsion Medical Systems, Feldkirchen, Germany) was used to monitor haemodynamic parameters. The pressure transducers were fixed with adhesive tape at the intersection of the right midaxillary line and fourth intercostal space. Transpulmonary thermodilution measurements were performed at the beginning of the study and after VLT by injecting 15-ml of cold saline into the central venous catheter, and the average of three consecutive measurements of the parameters at each time point was recorded, including transpulmonary thermodilution measurements of CI (CI_TPTD), global end-diastolic volume index (GEDVI), extravascular lung water index (ELWI), and global ejection fraction (GEF). After the initial thermodilution set of measurements, the parameters derived from the calibrated pulse contour analysis were recorded consecutively, including PPV, SVV, SVI, and CCI.

The change in percentage of SVI induced by TREND_SUPINE (ΔSVI_SUPINE-TREND) was measured as the difference between the maximum SVI at the Trendelenburg position and the SVI at baseline-1, divided by the SVI at baseline-1. Furthermore, the TREND_rTREND-induced percentage change in SVI (ΔSVI_rTREND-TREND) was calculated by dividing the difference between the maximum SVI in the Trendelenburg position and the minimum SVI in the reverse Trendelenburg position by the minimum SVI in the reverse Trendelenburg position. The PLR-induced percentage change of SVI (ΔSVI_PLR) was defined as the increase in SVI from the semi-recumbent position with the trunk elevated at 45° to the lower limbs elevated at 45° and the trunk in the supine position divided by the SVI in the semi-recumbent position with the trunk elevated at 45°. A positive volume response was defined as a VLT-induced SVI percentage change (ΔSVI_VLT) of ≥ 15% compared with baseline-4.

Data collection

Demographic information, Acute Physiology and Chronic Health Evaluation II (APACHE II) score, Sequential Organ Failure Assessment (SOFA) score, presence of shock and causes of shock, presence of ARDS, presence of acute kidney injury (AKI) and AKI classification, whether continuous renal replacement theory (CRRT) was performed, whether norepinephrine was administered and the dose of norepinephrine administered, respiratory mechanics parameters, and arterial blood gas analysis were collected at the time of patient inclusion. The diagnostic criteria for shock include clinical manifestations of tissue hypoperfusion, such as clammy or mottled skin, urine output less than 0.5 ml.kg⁻¹.h⁻¹, and altered levels of consciousness, either with or without hypotension (defined as arterial systolic pressure ≤ 90 mm Hg or a drop of ≥ 40 mm Hg from baseline blood pressure), or elevated blood lactate levels ≥ 2 mmol.L⁻¹ [17]. ARDS was diagnosed in accordance with the Berlin criteria [18], while AKI was diagnosed and classified following the Kidney Disease: Improving Global Outcomes guideline [19].

Statistical analysis

Values are presented as mean ± standard deviation for data that were normally distributed or median and interquartile range for data that were not normally distributed for continuous variables and number (%) for categorical variables. The Kolmogorov–Smirnov test was used to assess the normality and homogeneity of variance of all data. Comparisons between responders and non-responders were performed with Fisher’s exact test for categorical variables, and with a Student’s t-test, Mann–Whitney test, or analysis of variance (ANOVA) for continuous and ordinal variables when appropriate. Comparisons between two time points were performed using a paired Student’s t-test or a Wilcoxon rank-sum test. Hemodynamic parameters from baseline-1 to baseline-4 were compared using repeated measures ANOVA. The relationships between ΔSVI_SUPINE-TREND, ΔSVI_rTREND-TREND, ΔSVI_PLR, and ΔSVI_VLT were analyzed using linear regression. The predictive performance of ΔSVI_SUPINE-TREND, ΔSVI_rTREND-TREND, and ΔSVI_PLR for volume responsiveness were analyzed using the receiver operating characteristic curve (ROC) and expressed as the area under the curve (AUC) and 95% confidence interval (CI) of AUC. DeLong test was used to compare the area under the ROC curve [20]. The best cutoff value was obtained when the Youden index (sensitivity + specificity-1) was maximal [21]. The sensitivity, specificity, positive predictive value, negative predictive value, and 95% CI were calculated according to the best cut-off value. The gray zone of the optimal cutoff value was calculated according to the method presented by Cannesson et al. [22]. Results below the lower limit or above the upper limit of the gray zone were considered negative or positive, and results within the gray zone were considered inconclusive. SPSS 17.0.0 statistical package (SPSSInc, Chicago, USA) and MedCalc 20.217 software (Mariakerke, Belgium) were used for data processing and statistical analysis. Statistical significance was set at p < 0.05.

Results

Patients

During the study period, 46 patients who met the inclusion criteria were screened and 36 patients were eventually included in the analysis. Fifteen patients (42%) were volume-responsive after VLT. The patient inclusion process is illustrated in Fig. 2. The demographic and clinical characteristics, hemodynamic and respiratory parameters of the patients at the time of inclusion are shown in Tables 1 and 2.

Fig. 2.

Study inclusion flowchart

Table 1.

Demographic and clinical characteristics of patients at the time of inclusion

Characteristics	Overall population (n = 36)	Responders (n = 15)	Non-responders (n = 21)	p
Age (years)	67 ± 12	71 ± 8	64 ± 13	0.09
Male (%)	28 (78%)	13 (87%)	15 (71%)	0.42
BMI (kg.m−2)	23.8 ± 2.8	22.9 ± 3.4	24.5 ± 2.0	0.08
APACHE II score	27 ± 5	27 ± 3	27 ± 5	0.88
SOFA score	11 ± 4	10 ± 4	12 ± 4	0.23
Shock	29 (81%)	11 (73%)	18 (86%)	0.42
Causes of Shock
Septic shock	25 (69%)	10 (67%)	15 (71%)	1.00
Hypovolemic shock	4 (11%)	1 (7%)	3 (14%)	0.63
ARDS	5 (14%)	2 (13%)	3 (14%)	1.00
AKI	28 (78%)	11 (73%)	17 (81%)	0.69
AKI Class 1	7 (19%)	5 (33%)	2 (10%)	0.10
AKI Class 2	8 (22%)	3 (20%)	5 (24%)	1.00
AKI Class 3	11 (31%)	3 (20%)	9 (43%)	0.28
CRRT	15 (42%)	3 (20%)	12 (57%)	0.04

Values are expressed as the mean ± standard deviation or number (percentage)

BMI body mass index, APACHE II score Acute Physiology and Chronic Health Evaluation II score, SOFA Sequential Organ Failure Assessment, ARDS acute respiratory distress syndrome, AKI acute kidney injury, CRRT continuous renal replacement therapy

Table 2.

Hemodynamic and respiratory parameters of patients at the time of inclusion

Parameters	Overall population(n = 36)	Responders(n = 15)	Non-responders(n = 21)	p
Vasopressors administration	28 (78%)	10 (67%)	18 (86%)	0.24
Norepinephrine dose(μg.kg−1.min−1)	0.11 (0.02,0.39)	0.08 (0.00,0.21)	0.12 (0.04,0.43)	0.35
HR (min−1)	93 ± 21	91 ± 20	94 ± 22	0.63
MAP (mm Hg)	81 ± 15	78 ± 12	83 ± 17	0.35
CVP (mm Hg)	11 ± 5	9 ± 3	12 ± 5	0.04
GEF (%)	18 ± 6	17 ± 6	18 ± 6	0.52
GEDVI (ml.m−2)	653 ± 129	595 ± 106	694 ± 129	0.02
ELWI (ml.kg−1)	9.0 ± 3.1	8.0 ± 2.4	9.7 ± 3.4	0.12
CI (L.min−1.m−2)	2.56 ± 0.69	2.26 ± 0.56	2.77 ± 0.71	0.03
PPV (%)	10 ± 7	16 ± 7	7 ± 5	< 0.01
SVV (%)	12 ± 7	16 ± 7	8 ± 5	< 0.01
SVI (ml.m−2)	29 ± 10	26 ± 10	31 ± 10	0.20
Respiratory rate (min−1)	20	20 (20,20)	20 (20,20)	0.55
Tidal volume (ml.kg−1PBW)	9.1 ± 2.4	9.9 ± 2.3	8.4 ± 2.3	0.07
Resistance of the respiratory system (cm H2O.L−1.s−1)	12 (10,14)	11 (10,13)	11 (10,14)	0.42
Compliance of the respiratory system (ml.cm H2O−1)	54 ± 18	65 ± 16	46 ± 15	< 0.01
Pplat (cm H2O)	20 (20,20)	20 (18,20)	20 (20,21)	0.33
PEEP (cm H2O)	6 (6,7)	6 (6,6)	6 (6,8)	0.28
Driving pressure (cm H2O)	14 (13,14)	14 (13,14)	14 (13,14)	0.70
pH	7.42 ± 0.10	7.40 ± 0.11	7.43 ± 0.10	0.40
PCO2 (mm Hg)	33 ± 11	36 ± 10	32 ± 11	0.27
PaO2/FiO2 (mm Hg)	269 ± 86	286 ± 92	256 ± 82	0.31
Arterial lactate (mmol.L−1)	2.3 (1.5,6.0)	2.0 (1.2,4.1)	3.0 (1.6,7.6)	0.16

Values are expressed as the mean ± standard deviation, median [25th–75th percentile] or number (percentage)

HR heart rate, MAP mean arterial pressure, CVP central venous pressure, GEF global ejection fraction, GEDVI global end-diastolic volume index, ELWI extravascular lung water index, CI cardiac index, PPV pulse pressure variation, SVV stroke volume variation, SVI stroke volume index, PBW predicted body weight, Pplat plateau pressure, PEEP positive end-expiratory pressure, PaCO₂ partial pressure of arterial carbon dioxide, PaO₂ partial pressure of arterial oxygen, F_iO₂ inspired oxygen fraction

Hemodynamic monitoring

The entire study lasted 19 ± 2 min. There was no statistical difference in hemodynamic parameters from baseline-1 to baseline-4 (Table 3). After VLT, there was a significant increase in systolic arterial pressure (SAP), SVI, CCI, CI_TPTD, GEDVI, and CVP, and a significant decrease in HR, PPV, and SVV compared with baseline-1 (Table 3). During the study period, no adverse events were observed.

Table 3.

Hemodynamic parameters at each step

Parameters	Baseline-1	Trendelenburg position	Baseline-2	reverse Trendelenburg position	Trendelenburg position	Baseline-3	Semi-recumbent position	PLR	Baseline-4	Volume loading test
HR (min−1)	93 ± 22	92 ± 22	92 ± 22	92 ± 23	91 ± 22	92 ± 22	92 ± 23	92 ± 22	92 ± 22	89 ± 21b
SAP (mm Hg)	117 ± 20	128 ± 21a	117 ± 25	108 ± 28b	131 ± 20a	119 ± 24	119 ± 28	130 ± 23a	119 ± 25	133 ± 28a
SVI (ml.m−2)	29 ± 10	31 ± 11b	29 ± 11	27 ± 10a	31 ± 11a	29 ± 10	28 ± 11a	31 ± 11b	29 ± 10	33 ± 11a
CCI (L.min−1.m−2)	2.57 ± 0.71	2.73 ± 0.69	2.57 ± 0.69	2.38 ± 0.72a	2.72 ± 0.71b	2.56 ± 0.68	2.44 ± 0.70a	2.67 ± 0.69b	2.58 ± 0.70	2.79 ± 0.75a
CITPTD (L.min−1.m−2)	2.56 ± 0.69	–	–	–	–	–	–	–	–	2.77 ± 0.73a
GEDVI (ml.m−2)	650 ± 132	–	–	–	–	–	–	–	–	692 ± 132b
ELWI (ml.kg−1)	9.1 ± 3.2	–	–	–	–	–	–	–	–	9.1 ± 3.2
GEF (%)	18 ± 6	–	–	–	–	–	–	–	–	19 ± 7
PPV (%)	11 ± 7	8 ± 6b	11 ± 7	13 ± 8b	8 ± 6a	10 ± 7	10 ± 8	9 ± 6b	10 ± 6	7 ± 6a
SVV (%)	11 ± 7	8 ± 6a	12 ± 8	12 ± 7b	8 ± 6a	11 ± 8	10 ± 6	9 ± 6b	11 ± 7	8 ± 5a
CVP (mm Hg)	11 ± 4	14 ± 5a	11 ± 5	10 ± 4a	14 ± 5a	11 ± 5	12 ± 4	12 ± 4a	11 ± 4	13 ± 5a

Values are expressed as the mean ± standard deviation

HR heart rate, SAP systolic arterial pressure, SVI stroke volume index, CCI continuous cardiac index assessed by pulse contour analysis, CI_TPTD cardiac index assessed by transpulmonary thermodilution, GEDVI global end-diastolic volume index, ELWI extravascular lung water index, GEF global ejection fraction, PPV pulse pressure variation, SVV stroke volume variation, CVP central venous pressure

^ap < 0.001 vs. baseline-1; ^bp < 0.05 vs. baseline-1

Trendelenburg position initiated from the supine position

Compared to baseline-1, TREND_SUPINE resulted in a significant increase in SAP, SVI, CCI, and CVP, and a significant decrease in PPV and SVV, while HR remained unchanged (Table 3). The mean ∆SVI_SUPINE-TREND was significantly higher in responders than in non-responders (10% vs. 3%, p = 0.03; Fig. 3a). ΔSVI_SUPINE-TREND was significantly correlated with ΔSVI_VLT (R² = 0.14; Fig. 4a). The AUC of ΔSVI_SUPINE-TREND for predicting volume responsiveness was 0.78 (CI_95% 0.59–0.92, p < 0.001), with a sensitivity of 93% and specificity of 67% at a threshold of 4% (Table 4, Fig. 5a). The corresponding gray zone ranged from 0 to 6% and included 14% of patients (Table 4, Fig. 6a).

Fig. 3.

Individual values from ∆SVI_SUPINE-TREND, ∆SVI_rTREND-TREND, and ∆SVI_PLR to detect volume responsiveness in volume responders and non-responders. Open circles represent individual values. Red dashed lines display the optimal thresholds for each diagnostic test computed by receiver operating characteristic (ROC) curve analysis. The black horizontal line is the mean, while the upper and lower green lines represent the 95% confidence intervals for the mean. ∆SVI_SUPINE-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the supine position, ∆SVI_rTREND-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the reverse Trendelenburg position, ∆SVI_PLR: passive leg raising-induced stroke volume index percentage change

Fig. 4.

Linear regression between ∆SVI_SUPINE-TREND, ∆SVI_rTREND-TREND, ∆SVI_PLR, and ∆SVI_VLT. Solid and dashed lines indicate regression lines and their 95% confidence intervals, respectively. ∆SVI_SUPINE-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the supine position, ∆SVI_rTREND-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the reverse Trendelenburg position, ∆SVI_PLR: passive leg raising-induced stroke volume index percentage change, ∆SVI_VLT: Volume loading test-induced stroke volume index percentage change

Table 4.

Diagnostic performance of ∆SVI_SUPINE-TREND, ∆SVI_rTREND-TREND, and ∆SVI_PLR to predict volume responsiveness

Parameters	∆SVISUPINE-TREND	∆SVIrTREND-TREND	∆SVIPLR
AUROC (CI95%)	0.78 (0.59–0.92)	0.88 (0.75–0.96)	0.83 (0.61–0.95)
Youden index	0.6	0.63	0.59
Optimal threshold (CI95%)	4%	13%	11%
Grey zone of optimal threshold	0–6%	5–22%	6–17%
Patients in gray zone, [number (%)]	5 (14%)	15 (42%)	12 (33%)
Sensitivity (CI95%)	93% (68–100%)	87% (60–98%)	73% (45–92%)
Specificity (CI95%)	67% (43–85%)	76% (53–92%)	86% (64–97%)
Positive predictive value (CI95%)	67% (52–79%)	72% (55–85%)	79% (56–92%)
Negative predictive value (CI95%)	93% (67–99%)	89% (68–97%)	81% (65–91%)
Positive likelihood ratio (CI95%)	2.80 (1.51–5.20)	3.64 (1.65–8.02)	5.12 (1.72–15.29)
Negative likelihood ratio (CI95%)	0.10 (0.02–0.68)	0.18 (0.05–0.65)	0.31 (0.13–0.73)

AUROC area under receiver operating characteristic curve, CI_95% 95% confidence interval, ∆SVI_SUPINE-TREND trendelenburg position-induced stroke volume index percentage change initiated from the supine position, ∆SVI_rTREND-TREND trendelenburg position-induced stroke volume index percentage change initiated from the reverse Trendelenburg position, ∆SVI_PLR passive leg raising-induced stroke volume index percentage change

Fig. 5.

Receiver operating characteristic curves for ∆SVI_SUPINE-TREND, ∆SVI_rTREND-TREND, and ∆SVI_PLR to predict volume responsiveness. ∆SVI_SUPINE-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the supine position, ∆SVI_rTREND-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the reverse Trendelenburg position, ∆SVI_PLR: passive leg raising-induced stroke volume index percentage change

Fig. 6.

Sensitivity and specificity of ∆SVI_SUPINE-TREND, ∆SVI_rTREND-TREND, and ∆SVI_PLR to predict volume responsiveness and the gray zone of optimal thresholds. The gray zone is the 95% confidence interval for the best cutoff value. The gray zone is defined as a test result that is indeterminately positive or negative. ∆SVI_SUPINE-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the supine position, ∆SVI_rTREND-TREND: trendelenburg position-induced stroke volume index percentage change initiated from the reverse Trendelenburg position, ∆SVI_PLR: passive leg raising-induced stroke volume index percentage change

Trendelenburg position initiated from the reverse Trendelenburg position

Compared with baseline-1, the reverse Trendelenburg position resulted in a significant decrease in SAP, SVI, CCI, and CVP, and a significant increase in PPV and SVV, while HR remained unchanged (Table 3). Compared with baseline-1, TREND_rTREND resulted in a significant increase in SAP, SVI, CCI, and CVP, and a significant decrease in PPV and SVV, while the HR remained unchanged (Table 3). The mean ∆SVI_rTREND-TREND was significantly higher in responders compared to non-responders (27% vs. 8%, p < 0.001; Fig. 3b). Moreover, ∆SVI_rTREND-TREND was significantly correlated with ∆SVI_VLT (R² = 0.32; Fig. 4b). The AUC of ∆SVI_rTREND-TREND for predicting volume responsiveness was 0.88 (CI_95% 0.75–0.96, p < 0.001), with a sensitivity of 87% and specificity of 76% at a threshold of 13% (Table 4, Fig. 5b). The corresponding gray zone ranged from 5 to 22% and included 42% of patients (Table 4, Fig. 6b).

PLR

Compared to baseline-1, PLR resulted in a significant increase in SAP, SVI, CCI, and CVP, and a significant decrease in PPV and SVV, while HR remained unchanged (Table 3). The mean ∆SVI_PLR was significantly higher in responders than in non-responders (19% vs. 6%, p = 0.002; Fig. 3c). ΔSVI_PLR was significantly correlated with ΔSVI_VLT (R² = 0.43; Fig. 4c). ΔSVI_PLR predicted volume responsiveness with an AUC of 0.83 (CI_95% 0.61–0.95, p < 0.001), with a sensitivity of 73% and specificity of 86% at a threshold of 11% (Table 4, Fig. 5c). The corresponding gray zone ranged from 6 to 17% and included 33% of patients (Table 4, Fig. 6c).

Comparison ofTREND_SUPINE, TREND_rTREND, and PLR

ΔSVI_SUPINE-TREND was significantly lower than ΔSVI_rTREND-TREND and ΔSVI_PLR, 6% (CI_95% 3–9%), 16% (CI_95% 10–21%), and 11% (CI_95% 8–15%), respectively [ΔSVI_SUPINE-TREND vs. ΔSVI_{rSUPINE-TREND}, p < 0.001; ΔSVI_SUPINE-TREND vs. ΔSVI_PLR, p < 0.05]. There was no statistical difference in area under receiver operating characteristic curve (AUROC) for ΔSVI_SUPINE-TREND, ΔSVI_rTREND-TREND, and ΔSVI_PLR in predicting volume responsiveness (Fig. 5d).

Discussion

This study is the first to compare the performance of the Trendelenburg position and PLR in predicting volume responsiveness in mechanically ventilated patients in the ICU. The main findings are as follows: both Trendelenburg position (from supine and reverse Trendelenburg position)-induced and PLR-induced SVI percentage change can effectively predict volume responsiveness; ΔSVI induced by TREND_SUPINE, TREND_rTREND, and PLR had a similar ability to predict volume responsiveness.

PLR is a well-validated method for assessing volume responsiveness. Because its hemodynamic effects are reversible and independent of cardiopulmonary interaction, it is widely used in the ICU. PLR predicts volume responsiveness not only with high sensitivity and specificity, but also with very good positive and negative predictive values [6, 23]. In the Surviving Sepsis Campaign, PLR is recommended for the hemodynamic management of septic shock [24]. However, in several clinical scenarios (during surgery, prone position, pelvic and lower extremity fractures), PLR is not appropriate [10]. Therefore, there is an unmet need for a method to assess volume responsiveness that has the advantages of PLR while overcoming its limitations.

The Trendelenburg position, which mobilizes blood flow from the lower extremities and viscera to the heart due to gravity and temporarily increases cardiac preload, was used as an anti-shock position during the First World War [12, 25]. Yonis et al. [13] demonstrated that Trendelenburg position-induced changes in CI were highly accurate in predicting volume responsiveness in prone patients with ARDS. Ma et al. [26] showed that Trendelenburg position-induced changes in left ventricular outflow tract velocity–time integrals (LVOT ΔVTI) measured by transesophageal ultrasound were also of high value in predicting volume responsiveness in patients undergoing cardiac surgery. Luo et al. [14] applied Trendelenburg position-induced LVOT ΔVTI measured by transthoracic ultrasound to predict volume responsiveness in patients on V-A ECMO and achieved good results.

However, whether the Trendelenburg position has the same efficacy as PLR in predicting volume responsiveness has not yet been verified. The present study provides an answer to this question, finding no statistical difference in the AUROC for predicting volume responsiveness between the Trendelenburg position (from supine and reverse Trendelenburg position)-induced and PLR-induced changes in SVI. Therefore, the Trendelenburg position can be used as an alternative method of volume responsiveness assessment in certain situations where the PLR cannot be used.

PLR initiated from the semi-recumbent position induced a greater increase in preload than PLR initiated from the supine position and is the standard PLR method for predicting volume responsiveness [7, 27]. Similarly, the location of starting the Trendelenburg position is important. In the present study, it was demonstrated that ΔSVI_rTREND-TREND was significantly higher than ΔSVI_SUPINE-TREND, implying that TREND_rTREND mobilized more blood back to the heart than TREND_SUPINE. However, no differences in the efficacy of TREND_rTREND versus TREND_SUPINE-induced changes in SVI in predicting volume responsiveness were observed in the present study.

Furthermore, it was demonstrated that the threshold for TREND_SUPINE to predict volume responsiveness was a change in SVI of ≥ 4%, which is relatively small; therefore, detecting its effect requires real-time and precise measurement of stroke volume. The pulse contour analysis used in this study can detect minimal significant changes of up to 1–2% [28] and is a suitable technique for assessing this effect. Although echocardiography can provide continuous and real-time measurements of stroke volume, a change in VTI of more than 10% is required to be detectable on ultrasonography [29]. Therefore, ultrasound is not an appropriate method for assessing TREND_SUPINE-induced change in SVI. In this study, the threshold of the change in SVI percentage for TREND_rTREND to predict volume responsiveness was 13%, which is greater than the accuracy of ultrasound measurement. Therefore, in this case, ultrasound measurement of VTI can be used as an alternative method to the pulse contour analysis technique. This has been confirmed in previous literature [14, 26]. Because TREND_rTREND induces a greater percentage change in SVI which is easier to recognize and assess with ultrasound, TREND_rTREND is preferentially recommended for assessing volume responsiveness.

With the introduction of the "gray zone" concept for volume responsiveness assessment [30], particular care is required in applying the results of diagnostic tests for clinical decision making [31]. In the already published trials in which the Trendelenburg position predicted volume responsiveness, 30% [13], 29% [26], and 32% [14] of patients were in the gray zone, respectively, and in the present study, 14% and 42% of patients were in the gray zone, respectively, when TREND_SUPINE and TREND_rTREND predicted volume responsiveness. Therefore, the benefits and risks of increasing or limiting the fluid load in patients near the cutoff value of diagnostic testing still needs to be weighed individually in clinical practice.

Limitations

Our study has several limitations. First, this was a small-sample study, and a small sample size may result in a high proportion of patients in the gray zone and affect the comparison of the AUROC for diagnostic tests. Therefore, studies with larger sample sizes are required to validate the results of this study. Second, in this study, a fluid bolus of 250 ml was administered for fluid challenge, and there was concern that this did not induce a sufficient increase in preload. A fluid challenge volume that is too small is not sufficient to stretch the ventricles and thus produce a sufficient increase in CO. In this sense, an increase in CVP owing to a fluid challenge identifies an appropriate fluid challenge. For example, an increase in CVP of 2 cmH₂O is evidence of proper right ventricular stretch [32]. In addition, the appropriate amount of fluid challenge needs to balance the benefit of increased CO with the risk of elevated filling pressure. If the increase in CVP after fluid challenge is > 2 cmH₂O, the risk of fluid challenge is increased [2]. In this study, the mean increase in CVP after 250 ml of 4% albumin fluid challenge was 2 cmH₂O, implying that fluid challenge induced an appropriate increase in cardiac preload without increasing the risk of fluid challenge. Third, a meta-analysis showed that the volume of fluid challenge used to determine the change in CO significantly affects the predictive performance of PLR [9]. However, whether it affects the performance of the Trendelenburg position in predicting volume responsiveness requires further research. Fourth, to avoid too much fluid infusion, the CO of pulse contour monitoring was not recalibrated by thermodilution at each step of the protocol. However, in this study, the entire course of the trial was completed in an average of 19 min, which reduced the influence of factors other than postural changes on hemodynamics. Fifth, awake, non-mechanically ventilated, and arrhythmic patients were not included in this study, and whether the findings apply to these populations requires further verification. Sixth, different angled Trendelenburg positions produce different volumes of autologous infusion [12]. The Trendelenburg position used in this study was a 15-degree downward tilt of the bed. Therefore, the results of this study cannot be extrapolated to other angled Trendelenburg positions. Seventh, this study required an electric bed for position conversion, and application of the results of this study may be limited in resource-limited areas.

Conclusion

The study showed that the Trendelenburg position (TREND_SUPINE and TREND_rTREND)-induced and PLR-induced SVI changes were comparable in their ability to predict volume responsiveness; therefore, the Trendelenburg position may be a reasonable alternative to PLR for predicting volume responsiveness in certain clinical scenarios. Because TREND_rTREND induces greater SVI changes than TREND_SUPINE, TREND_rTREND is preferred for predicting volume responsiveness.

Abbreviations

ICU

Intensive care unit

PLR

Passive leg raising

ARDS

Acute respiratory distress syndrome

V-A ECMO

Veno-arterial extracorporeal membrane

CO

Cardiac output

SVI

Stroke volume index

ΔSVI

Percentage change in stroke volume index

PiCCO

Pulse indicator continuous cardiac output

VLT

Volume loading test

RASS

Richmond agitation-sedation scale

HR

Heart rate

BP

Blood pressure

CCI

Continuous cardiac index

PPV

Pulse pressure variation

SVV

Stroke volume variation

CVP

Central venous pressure

CI_TPTD

Transpulmonary thermodilution measurements of CI

GEDVI

Global end-diastolic volume index

ELWI

Extravascular lung water index

GEF

Global ejection fraction

ΔSVI_SUPINE-TREND

Trendelenburg position-induced stroke volume index percentage change initiated from the supine position

ΔSVI_rTREND-TREND

Trendelenburg position-induced stroke volume index percentage change initiated from the reverse Trendelenburg position

ΔSVI_PLR

Passive leg raising-induced stroke volume index percentage change

ΔSVI_VLT

Volume loading test-induced stroke volume index percentage change

APACHE II

Acute physiology and chronic health evaluation II

SOFA

Sequential organ failure assessment

AKI

Acute kidney injury

CRRT

Continuous renal replacement therapy

ANOVA

Analysis of variance

ROC

Receiver operating characteristic curve

AUC

Area under the curve

CI_95%

95% Confidence interval

SAP

Systolic arterial pressure

AUROC

Area under the receiver operating characteristic curve

LVOT ΔVTI

Changes in left ventricular outflow tract velocity–time integrals

Author contributions

ZYW made significant contributions to study design, data collection, data analysis and interpretation, and manuscript drafting; JZ, JZ, YXW, and SYZ made significant contributions to data collection and data analysis; CFY made significant contributions to data analysis; and XJG and LX made significant contributions to study design, data interpretation, and manuscript revision.All authors read and approved the final version of the manuscript.

Funding

Tianjin Science and Technology Project (21YCFBJC01200), Tianjin Key Medical Discipline(Specialty)construction Project (TJYXZDXK-035A), Research Project on the Integration of Traditional Chinese Medicine and Western Medicine by Tianjin Health Commission (2023221).

Availability of data and materials

The data that support the findings of this study are available on request from the corresponding author, ZYW, upon reasonable request.

Declarations

Ethics approval and consent to participate

The present study was approved by the Institutional Review Board (Ethics committee of Tianjin Third Central Hospital, Tianjin, China. Ethics Approval Number: IRB2019-037-01). Written consent from the patients’ closest relatives was required for inclusion.

Consent for publication

The manuscript has been read and its submission approved by all coauthors. Patients were prospectively included after informed consent from the patient’s next of kin.

Competing interests

The authors declare no competing interests.