Passive leg raising and microvascular skin blood flow to predict peripheral tissue perfusion fluid responsiveness

Abstract

Objective

To assess whether skin blood flow (SBF) monitoring combined with passive leg raising (PLR) can predict microvascular fluid responsiveness in septic patients.

Design

Prospective observational study.

Setting

Single-center, 18-bed medical ICU in a tertiary university hospital in Paris, France.

Patients

Adult patients with sepsis requiring intravenous fluid administration.

Interventions

Patients underwent a standardized PLR maneuver followed by a 500 mL saline fluid administration. Peripheral SBF was continuously monitored by fingertip laser Doppler flowmetry.

Measurements and main results

Of 37 patients included, 27 (73%) were classified as fluid responders, defined by a > 15% increase in SBF after volume expansion (ΔSBF-VE). In responders, SBF increased significantly during PLR (ΔSBF-PLR 40% [21–105]), while no significant changes were observed in non-responders. SBF variations induced by PLR (ΔSBF-PLR) strongly predicted fluid responsiveness with an AUROC of 0.95 [0.86–1.00] (P < 0.001). A ΔSBF-PLR threshold of > 6% identified responders with an 96 [80–100] % sensitivity and 90 [59–100] % specificity. Positive predictive value was 96 [80–100] % and negative predictive value was 91 [59–100]. Changes in SBF did not correlate with changes in cardiac output after volume expansion (R² =0.04, P = 0.28).

Conclusions

In septic patients, PLR-induced changes in SBF reliably predict peripheral microvascular responsiveness to a subsequent volume expansion. This simple, non-invasive approach may facilitate personalized fluid strategies aimed at optimizing microvascular tissue perfusion.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-026-05852-x.

Key points

Question

Can skin blood flow (SBF) changes induced by passive leg raising (PLR) predict microvascular fluid responsiveness in septic patients?

Findings

In this prospective observational study, PLR-induced SBF changes accurately predicted fluid responsiveness (AUROC of 0.95 [0.86–1.00]) defined by a > 15% increase in SBF after volume expansion.

Meaning

PLR-induced SBF responsiveness is a simple, noninvasive method to predict peripheral microvascular fluid responsiveness in sepsis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-026-05852-x.

Introduction

Sepsis remains a major global health issue, and improving outcomes requires more effective strategies for resuscitation [1]. In patients with sepsis, peripheral tissue hypoperfusion is strongly associated with poor outcome and correlates with organ failure [2]. Fluid resuscitation is a cornerstone of therapeutic management in sepsis, aiming to restore optimal volemia and ultimately improve tissue perfusion [3]. However, the therapeutic benefit of fluids depends on appropriate dosing, insufficient fluid may lead to ongoing organ hypoperfusion, while excessive fluid can result in overload and deleterious tissue congestion. Both under-resuscitation and fluid overload have been independently associated with increased mortality in critically ill patients [4, 5]. Accordingly, the Surviving Sepsis Campaign and the European Society of Intensive Care Medicine recommend personalizing fluid therapy by assessing fluid responsiveness, thereby limiting fluid administration to patients most likely to benefit from it [6, 7]. Among available dynamic tests to evaluate fluid responsiveness, passive leg raising (PLR) is one of the most widely used [8]. By shifting approximately 300 mL of blood from the lower extremities toward the central circulation, PLR mimics the effects of volume expansion without actually administering fluid. This reversible “fluid challenge” has been validated as a predictor of cardiac output response to volume expansion [9]. However, in sepsis, loss of hemodynamic coherence is common: improvements in macrohemodynamic parameters, such as cardiac output, do not necessarily translate into improved organ perfusion [10, 11], . In such cases, an increase in cardiac output may not improve the microvascular perfusion. These observations underscore the potential value of incorporating microcirculatory assessment into fluid resuscitation strategies. Skin blood flow (SBF) monitoring, allows non-invasive, real-time, quantitative measurement of microcirculatory perfusion using the laser doppler technology [12]. This tool has been gaining interest as recent studies have demonstrated significant associations between SBF and established clinical markers of tissue perfusion, such as capillary refill time (CRT), lactate level, organ failure severity and mortality [13, 14].

This study aimed to evaluate whether the combination of passive leg raising with SBF monitoring, can predict microvascular fluid responsiveness in septic patients.

Materials and methods

Study design and setting

We conducted a 3-month prospective observational study in an 18-bed ICU tertiary teaching hospital in Paris. This observational study protocol was approved by the ethical committee, Comité de Protection des Personnes (Comité de protection des personnes Ouest II N° 2023-A02046-39). All patients or their legal representatives were informed of the study and gave oral consent for participation. This study is reported according to the STROBE guidelines for observational studies.

Patients

All adult patients (≥ 18 years) presenting with sepsis for whom the attending physician had decided to perform a fluid expansion were eligible for inclusion. Fluid infusion was decided exclusively by the clinician in charge of the patient and the decision was based on non-standardized parameters such as CRT, mottling, subjective assessment of peripheral temperature, urinary output, arterial hypotension and lactate levels. Sepsis was defined according to the third international consensus definition [2]. Exclusion criteria were lower limb amputation, suspected intra-abdominal hypertension or the use of compression stockings, as these conditions may impair the diagnostic accuracy of the PLR test [15]. Patients with severe agitation preventing reliable SBF measurement, as well as those with a history of Raynaud’s disease that may alter peripheral perfusion, were also excluded.

Study protocol and measurements

Baseline demographic and clinical data were collected at ICU admission including age, sex, reason for admission, Sequential Organ Failure Assessment (SOFA) score [16], and Simplified Acute Physiology score II (SAPS II) [17].

The study protocol consisted of four sequential measurement time points: T1: Baseline (semi-recumbent position, 45°); T2: After 1 min of PLR; T3: Return to baseline (semi-recumbent); T4: Immediately after volume expansion (500 mL Saline infused over 15 min). In accordance with established recommendations, PLR was performed by lowering the patient’s upper body from a 45° semi-recumbent to a supine position, while lower limbs were raised to a 45° angle [18]. At each time point, variables reflecting macro-hemodynamics and organ perfusion were collected. Macro-hemodynamic assessment included mean arterial pressure (MAP), heart rate (HR), cardiac output (CO) measured by transthoracic echocardiography and norepinephrine dosage (if applicable). Markers of tissue perfusion included index CRT and fingertip SBF. As previously described [13], SBF was evaluated using a skin laser doppler device (PeriFlux System 5000; Perimed, Jarfalla, Sweden). The emitted laser beam has a wavelength of 780 nm, allowing evaluation at 0.5 to 1 mm depth in the skin. The frequency shifts of the back-scattered light induced by moving red blood cells is proportional to their velocity, providing a continuous quantitative measurement of SBF in perfusion units (PUs).

SBF measurement was standardized by attaching the probe with a double-sided tape on the palmar surface of the index finger on the hand opposite to the arterial line or blood pressure cuff. To avoid confounders, no changes in therapeutic interventions were made between measurements, and ICU room temperature was controlled and maintained constant. SBF was measured continuously, and averaged over a one-minute period at each time point. To avoid interference or displacement of the SBF probe, CRT and SBF were measured on contralateral sides. In a subset of 10 patients, 3 repeated SBF measurements were performed to assess intra-observer reproducibility.

Endpoints

The primary endpoint was to determine the diagnostic ability of the change in SBF during passive leg raising (ΔSBF-PLR) to predict fluid responsiveness as previously. Fluid responsiveness was defined as a > 15% increase in SBF following volume expansion (ΔSBF-VE > 15%), assuming that a 15% variation is clinically relevant [13, 14]. Secondary endpoints included evaluation of the relationship between ΔSBF-VE, ΔSBF-PLR, and macrocirculatory variables.

Statistical analysis

Based on a previous work by Jacquet-Lagrèze et al. [19]. using Obuchowski’s methodology [20], for an expected AUROC of at least 0.8, an alpha risk of 0.05 and a power of 0.9, we estimated the target sample size to 40 patients.

Patients’ characteristics were summarized as medians with interquartile ranges (25–75th percentiles) or counts and percentages as appropriate. Group comparisons were performed using Fisher’s test for categorical variables and Mann–Whitney test for unpaired continuous variables. Comparisons between paired time points (T1 vs. T2 and T3 vs. T4) were conducted using a paired Wilcoxon signed-rank test. Changes in SBF were expressed as relative variations from baseline calculated as follows: ΔSBF = (Value_post – Value_baseline) / Value_baseline, ΔSBF-PLR = (T2 – T1) / T1 and ΔSBF-VE = (T4 – T3) / T3. To evaluate the diagnostic performance of ΔSBF-PLR for predicting fluid responsiveness (ΔSBF-VE > 15%) Receiver operating characteristic (ROC) curves were constructed. Results are reported as area under the ROC curve (AUC) with corresponding standard deviation (SD) and 95% confidence intervals (CIs). Sensitivity and specificity values were also calculated for the optimal threshold, determined by the Youden Index (sensitivity + specificity – 1). Correlations between variables were assessed using Pearson’s formula. Statistical significance was set at a two tailed p value < 0.05. All analyses were made using Graphpad Prism softwares (Graphpad Softwares, La Joya, CA).

Results

Intra-rater SBF reproducibility

In the subset of 10 patients, SBF was averaged over a 1-minute recording three times at 10-minute intervals. Intra-rater concordance was > 0.99, (p < 0.0001). The coefficient of variation for SBF was below 2%, indicating that a change greater than 4% can be considered real.

Studied population

During the 3-month study period, 45 consecutive septic patients were eligible. Eight patients were excluded, 5 patients because of intra-abdominal hypertension which impaired hemodynamic effects of PLR and 3 because of agitation impeding placement of the SBF measuring device. Baseline characteristics of the 37 patients included are summarized in Table 1. The principal sources of infections were pulmonary (N = 17, 46%) and abdominal (N = 7, 19%). The median time from admission to inclusion was 13 [4–21] hours. Seventeen patients (46%) patients underwent mechanical ventilation, and 12 (32%) received norepinephrine with a median dose of 0.35 [0.12–0.46] µg/kg/min at inclusion. The median SOFA score at inclusion was 8 [4–10] and the median SAPSII was 43 [33–62]. The 28-day mortality rate was 21%. At baseline, 17 patients (46%) had an index CRT greater than 3 s and 26 patients (70%) had a SBF under 100 PU.

Table 1.

Baseline characteristics of studied population

Parameters	Total (n = 37)	Responders (n = 27)	Non-responders (n = 10)	p-value
Age, years	65 (49–76)	58 (48–74)	74 (63–75)	0.10
Body mass index, kg/m²	22.4 (19.3–27.9)	24 (20–28)	21 (19–24)	0.54
Sex, female, n (%)	14 (38)	10 (37)	4 (40)	0.87
SAPS II	43 (33–62)	43 (30–65)	43 (38–53)	0.90
SOFA	8 (4–10)	8 (4–10)	8 (5–11)	0.30
Sepsis cause, n (%)
Lung	17 (46)	12 (44)	5 (50)	0.90
Abdomen	7 (19)	5 (19)	2 (20)	0.64
Urinary tract	4 (11)	1 (4)	3 (30)	0.05
Skin	4 (11)	1 (4)	3 (30)	0.05
Others	5 (14)	3 (11)	2 (20)	0.59
Time from admission to inclusion (hours)	13 (4–21)	12 (2–18)	18 (15–45)	0.05
Fluid administration before inclusion Liters	1.2 (0.6–2.0)	1.5 (0.7–2.0)	1.3 (1.0–1.8)	0.81
Norepinephrine N (%) Dose, µg/kg/min	12 (32) 0.35 (0.12–0.46)	8 (30) 0.35 (0.10–0.40)	4 (40) 0.49 (0.17–0.82)	0.55 0.35
Mechanical ventilation, n (%)	17 (46)	10 (37)	7 (70)	0.07
Mean arterial pressure, mmHg	75 (70–85)	76 (70–88)	74 (68–79)	0.49
Heart rate, beats/min	106 (87–121)	107 (92–126)	102 (81–108)	0.29
Cardiac output, L/min	5.0 (3.7–6.3)	4.4 (3.5–5.9)	6 (3.3–6.6)	0.31
Arterial Lactate, mmol/l	2.1 (1.4–4.9)	2.3 (1.5–5.1)	1.2 (0.9–1.2)	0.004
Index capillary refill time, seconds	2.7 (1.7–4.1)	2.8 (2–4.1)	2.5 (1.3–3.7)	0.55
Skin blood flow, Perfusion Units	44 (15–151)	35 (14–100)	90 (23–169)	0.32

CRT, Capillary Refill Time; SOFA, Sequential Organ Failure Assessment; SAPSII, Simplified Acute Physiology Score

Following fluid administration, 27 patients (73%) exhibited an increase in SBF > 15% and were classified as responders; the remaining 10 patients (27%) were considered non-responders.

Changes in SBF in responders and non-responders

Macrohemodynamic and tissue perfusion variables across study time points are presented in Table 2. In responders, fingertip SBF significantly increased from 34 [14–100] PU at baseline (T1) to 55 [33–159] PU, 1 min after PLR (T2) (p < 0.0001), corresponding to a 40% [21–105] increase. After PLR discontinuation (T3), SBF returned to 35 [17–149] PU and significantly increased to 68 [43–211] PU after volume expansion (T4) (p < 0.0001) representing a 97% [51–166] increase (Table 2; Fig. 1). In contrast, non-responders’ fingertip SBF exhibited no significant SBF changes after PLR or volume expansion (Table 2; Fig. 1).

Table 2.

Changes in hemodynamic parameters in responders and non-responders

SBF response	Parameters	Baseline (T1)	PLR (T2)	P value	Baseline (T3)	VE (T4)	P value
Responders	SBF, PU	34 (14–100)	55 (33–159)	< 0.0001	35 (17–149)	68 (43–211)	< 0.0001
	CRT, s	2.8 (2.1–4.1)	2.0 (1.6–2.8)	< 0.0001	2.7 (1.9–3.6)	2.2 (1.2–2.6)	< 0.0001
	CO, L/min	4.4 (3.5–5.9)	5.2 (4.2–6.7)	0.0008	5.1 (3.6–6.1)	5.1 (4.1–6.4)	0.03
	HR, bpm	107 (92–126)	107 (87–125)	0.83	107 (90–129)	108 (90–123)	0.06
	MAP, mmHg	76 (70–88)	83 (66–92)	0.09	80 (70–88)	77 (69–93)	> 0.99
Non Responders	SBF, PU	90 (23–169)	91 (19–157)	0.10	92 (21–124)	85 (22–119)	0.16
	CRT, sec	2.5 (1.3–3.7)	2.8 (1.9–3.9)	0.08	2.1 (1.2–3.9)	2.0 (1.5–3.3)	0.48
	CO, L/min	6.3 (6–7.1)	7 (7–7.4)	0.19	6 (5.8–6.7)	7.2 (7–7.8)	0.07
	HR, bpm	105 (81–121)	99 (77–107)	0.49	96 (77–106)	93 (73–107)	0.25
	MAP, mmHg	74 (68–79)	81 (68–92)	0.15	83 (68–91)	82 (70–86)	0.70

SBF, skin blood flow; CRT, Capillary refill time; CO, Cardiac output; HR, Hear rate; MAP, Mean arterial pressure

Fig. 1.

Representative kinetic of fingertip SBF across the four study time-points in non-responders (A) and responders (B)

Prediction of microvascular fluid responsiveness

PLR-induced variations in SBF (ΔSBF-PLR) accurately predicted fluid responsiveness, with an AUROC of 0.95 [0.86–1.00] (p < 0.001) (Fig. 2). A threshold ΔSBF-PLR > 6% predicted fluid responsiveness with a sensitivity of 96% [80–100] and a specificity of 90% [59–100]. Positive predictive value (PPV) was 96 [80–100] %, negative predictive value (NPV) was 91 [59–100]. A more clinically relevant threshold at 10% predicted fluid responsiveness with a sensitivity of 88%, a specificity of 90%, a PPV of 96% and a VPN negative of 77%.

Fig. 2.

AUROC of PLR prediction of SBF fluid responsiveness

Across the entire cohort, ΔSBF-PLR strongly correlated with ΔSBF after volume expansion (ΔSBF-VE) (R² = 0.60, p < 0.0001) (Fig. 3). This correlation was consistent among patients with or without mechanical ventilation (R² = 0.63, p < 0.0001 and R² = 0.60, p = 0.0003, respectively) (Supplementary Fig. 1a), as well as in patients receiving norepinephrine or not (R² = 0.71, p = 0.0006 and R² = 0.61 p < 0.0001, respectively) (Supplementary Fig. 1b). A stronger correlation was observed in patients with baseline SBF < 100 (R² = 0.66, p < 0.0001) compared to those with baseline SBF > 100 (R² = 0.54, p < 0.01) (Supplementary Fig. 1c).

Fig. 3.

Relationship between ΔSBF-PLR and ΔSBF-VE

Correlation with cardiac output

Cardiac output was measured in 27 patients using echocardiography. No significant correlation was found between ΔSBF and ΔCO following volume expansion (R²=0.04, p = 0.28) (Supplementary Fig. 2). Among the seven SBF non-responders, three (43%) were CO responders as defined by a ΔCO-VE > 15% [9]. Conversely, among the 20 SBF responders, 10 (50%) were CO non-responders (Supplementary Fig. 3). Overall, CO responsiveness predicted SBF responsiveness with a sensitivity of 50% and a specificity of 57%.

Correlation with CRT

At baseline, CRT and SBF were strongly correlated (R² = 0.84, p < 0.001). Changes in CRT and SBF also correlated significantly during both PLR and volume expansion (R² = 0.41, p < 0.0001 and R² = 0.42, p < 0.0001, respectively) (Supplementary Fig. 4).

Discussion

In this prospective study, we found that changes in fingertip SBF induced by PLR accurately predicted SBF responsiveness to a subsequent volume expansion. The optimal threshold identified using the Youden Index was an increase in SBF > 6% during PLR. We also found a significant correlation between PLR induced and volume expansion induced changes in SBF, whereas no significant correlation was observed between variations in SBF and macrohemodynamic parameters, particularly cardiac output.

Fluid administration is one of the most commonly prescribed treatments in the ICU, used in various conditions, especially in sepsis and shock [21]. While early fluid resuscitation is often beneficial for hemodynamic optimization [22], excessive administration can be detrimental by impairing vascular integrity and promoting edema, thereby contributing to organ dysfunction [23, 24]. Therefore, individualized fluid management strategies are crucial, and fluid responsiveness assessment is recommended whenever feasible [25]. We chose PLR as it is an easy-to-use, reversible fluid challenge that reliably predicts fluid responsiveness regardless of ventilation mode or underlying cardiac rhythm [26].

Traditionally, fluid responsiveness has been assessed through changes in cardiac output. However, accumulating evidence suggest that increases in cardiac output may not always translate into improved tissue perfusion due to the phenomenon of loss of hemodynamic coherence [27]. In our study, we chose to evaluate fluid responsiveness with a tissue perfusion readout, by using SBF. Previous studies have demonstrated that SBF is a robust method for evaluating tissue perfusion, showing strong correlation with capillary refill time [13] and significant association with both organ dysfunction severity and mortality in patients with circulatory shock [14]. Moreover SBF and its variations have been associated with outcome in circulatory failure [13, 14]. Finally, Mongkolpun suggested that fluid responsiveness defined by SBF increase was more closely associated to an increase in oxygen consumption than a cardiac output-based definition alone [28].

In this study, fluid responsiveness was defined as an increase of at least 15% in SBF following volume expansion, a threshold selected to exceed the least significant change in SBF and ensure clinical relevance [13, 14].

To ensure reproducibility, SBF measurement was standardized, and PLR was performed as recommended by lowering the patient’s upper body from a 45° semi-recumbent to a supine position, while lower limbs were raised to a 45° angle [18]. A significant but imperfect correlation was observed between SBF changes induced by PLR and those induced by volume expansion (R² = 0.60, p < 0.0001). Several factors may explain this imperfect correlation. First PLR induces an estimated auto-transfusion of 300 mL, whereas volume expansion delivers 500 mL of saline solution. Second, differences in blood between auto-transfused blood during PLR and crystalloid infusion during volume expansion (such as osmolarity, oncotic pressure and viscosity) may influence microvascular responses [29]. Third, the effects of PLR are considered immediate and assessed within one minute whereas volume expansion is evaluated after full infusion, usually lasting approximately 15 to 30 min.

In our study, no correlation was observed between ΔSBF and ΔCO. Notably, some patients were classified as CO responders but as SBF non-responders. This finding reinforces the concept that macrohemodynamic improvements do not necessarily guarantee enhanced tissue perfusion, thereby illustrating the concept of loss of hemodynamic coherence. It is important to acknowledge however that CO was assessed using echocardiography, which was not feasible in all patients and is subject to operator variability. Furthermore, the transient nature of PLR-induced CO variations (lasting only 60–90 s) may limit the ability of echocardiography to reliably capture these effects.

Our findings are consistent with a previous work evaluating PLR to predict peripheral perfusion fluid responsiveness using CRT [19]. However, CRT assessment in that study was relatively time-consuming and not easily applicable in routine clinical practice. In addition, CRT carries several limitations including operator dependence and susceptibility to factors such as skin pigmentation and ambient light [30–32]. In contrast, SBF may offer potential advantages as it provides a quantitative, continuous, real-time assessment of tissue perfusion that is unaffected by these factors [33, 34].

This study has several limitations. First, it is a monocentric study, with a unique case mix, and fluid expansion decision was left to the treating physician. Thus temporal and or physiological differences between patients could influence microvascular responsiveness and limit generalizability. Second, although our method was accurate in predicting peripheral tissue fluid responsiveness, its impact on clinical outcomes and its feasibility in daily practice remain to be determined. Third, we explored global microvascular blood flow, but we did not investigate the specific mechanisms underlying alterations in skin perfusion, such as endothelial or smooth muscle cell dysfunction, which can be assessed using a thermal challenge. Finally, no device is currently available for routine bedside measurement of SBF, which limits the immediate applicability of this approach. Nevertheless, ongoing technological advances may facilitate the development of practical, real-time SBF monitoring tools, potentially enabling wider integration of tissue perfusion guided fluid management in critical care. Taken together, these finding suggest that combining PLR with SBF measurement could be valuable for personalizing fluid resuscitation in septic patients.

Conclusion

In this prospective study, PLR-induced changes in fingertip SBF reliably predict the subsequent SBF response to volume expansion in patients with sepsis. These findings highlight the potential of combining SBF monitoring with PLR as a simple, noninvasive bedside approach to guide fluid resuscitation. This strategy may contribute to the development of personalized, tissue perfusion–oriented fluid management in critically ill patients with sepsis.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Study concept and design HAO. Acquisitions of data AM, TU, JB, LR, VB, LM and HAO. Data analysis, figures and statistics AM, PYB and HAO. Drafting of the manuscript AM and HAO. Critical revision of the manuscript, all the authors. All authors read and approved the final manuscript.

Funding

None.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The protocol was approved by a national IRB (Comité de protection des personnes Ouest II N° 2023-A02046-39). All patients were informed of the study and gave oral consent for participation.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.