Secondary Logo

Journal Logo


Accuracy of stroke volume variation compared with pleth variability index to predict fluid responsiveness in mechanically ventilated patients undergoing major surgery

Zimmermann, Markus; Feibicke, Thomas; Keyl, Cornelius; Prasser, Christopher; Moritz, Stefan; Graf, Bernhard M; Wiesenack, Christoph

Author Information
European Journal of Anaesthesiology: June 2010 - Volume 27 - Issue 6 - p 555-561
doi: 10.1097/EJA.0b013e328335fbd1
  • Free



Accurate assessment of a patient's volume status is an important goal for the anaesthetist in the operating theatre to achieve haemodynamic stability and adequate tissue oxygenation. As static indices of cardiac preload such as central venous pressure (CVP) and pulmonary artery wedge pressure are of little help for decisions regarding volume replacement,1,2 dynamic variables such as pulse pressure variation (PPV) and stroke volume variation (SVV) are increasingly used to detect the cyclic fluctuation of the arterial pressure wave in the mechanically ventilated patient in order to predict fluid responsiveness.3–5 Goal-directed intraoperative volume optimization using PPV as a haemodynamic endpoint has recently been shown to reduce the duration of mechanical ventilation and hospital stay, as well as postoperative complications in high-risk surgical patients.6 This accentuates the need for reliable indicators of fluid responsiveness to avoid needless or even deleterious volume replacement, which is associated with increased morbidity and mortality in mechanically ventilated patients.7 However, most dynamic variables assessing fluid responsiveness are invasive, technically challenging or require additional catheters.

The recently introduced FloTrac/Vigileo system (Edwards Lifesciences, Irvine, California, USA) requires access to a standard arterial catheter to enable less invasive assessment of arterial pressure-based cardiac output (CO), stroke volume and SVV without an external calibration mode. Accuracy of CO measurement with this new technique using the latest version of the software (V 1.10) shows sufficient agreement with the intermittent thermodilution method involving the pulmonary artery catheter (PAC).8–10 In addition to this, recent studies demonstrated the value of FloTrac-derived SVV in predicting fluid responsiveness in various clinical settings.11–13

Respiratory variations in the pulse oximeter plethysmographic waveform amplitude (ΔPOP) have been demonstrated to predict fluid responsiveness, but this variable could not be measured continuously at the bedside until now.14 The new Masimo Radical-7 monitor (Masimo Corp., Irvine, California, USA) provides a novel algorithm for automated, noninvasive and continuous estimation of ΔPOP by calculating pleth variability index (PVI). However, there have been few clinical investigations into the ability of this perioperative variable to predict fluid responsiveness.15–17 Moreover, accuracy of SVV and PVI to adequately predict fluid responsiveness has never been compared in a clinical trial.

Therefore, the purpose of the study was to compare the accuracy of less invasively assessed arterial pressure-based SVV and respiratory variations in the pulse oximeter plethysmographic (POP) waveform amplitude by noninvasively calculated PVI with CVP to predict the volume-induced response of stroke volume index (SVI) in normoventilated patients undergoing major surgery, before being able to investigate the effect of SVV-guided or PVI-guided intraoperative volume optimization in a forthcoming study.


After obtaining approval from the Local Ethics Committee and with written informed consent, we studied 20 patients (13 men) undergoing elective major abdominal surgery, who, due to the severity of their illness, required both arterial and CVP monitoring. Patients with valvular heart disease, intracardiac shunts, regional myocardial asynchrony, peripheral vascular disease, preoperative dysrhythmias and an ejection fraction of less than 30% were excluded from the study.

Anaesthesia was induced with fentanyl (3 μg kg−1) followed by propofol until loss of consciousness, and rocuronium (0.6 mg kg−1). Anaesthesia was maintained using 1.2% end-expiratory sevoflurane, supplemented with bolus doses of fentanyl (up to 3 μg kg−1) and rocuronium (0.2 mg kg−1). Volume-controlled ventilation with a positive end-expiratory pressure of 5 mmHg and a constant tidal volume of 7 ml kg−1 to an end-tidal pCO2 of 35–40 mmHg was maintained at 0.5 FiO2 throughout the study.

After induction of anaesthesia, all patients received a radial arterial line (20G; Becton Dickinson, Franklin Lakes, New Jersey, USA) for continuous monitoring of arterial blood pressure (Siemens monitor SC 9000; Siemens AG, Erlangen, Germany) and a triple lumen 20 cm 7.5 Fr central venous catheter (Arrow International Inc., Reading, Pennsylvania, USA), which was inserted into the right internal jugular vein for continuous assessment of CVP. The FloTrac sensor was attached to the arterial line and connected to the Vigileo monitor. Following initiation of the Vigileo monitor by entering the patient's age, sex, height and weight, the system computes stroke volume from the patient's arterial pressure signal and displays CO and SVV continuously.

The FloTrac/Vigileo system has been previously described in detail18 and involves the calculation of stroke volume based on the proven relationship between pulse pressure (PP) and stroke volume.19 The device assesses the arterial pulse waveform at a sampling rate of 100 Hz over a 20 s period, which generates approximately 2000 data points and calculates the standard deviation (SDAP) of each measured beat to provide a robust assessment of key PP characteristics. To facilitate CO assessment by arterial pulse waveform analysis without an external calibration mode, the FloTrac algorithm used (V 1.14) estimates individual arterial compliance according to Langewouter's five-component model20 and continuously compensates for changes in vascular tone by detecting characteristic alterations in the arterial pressure waveform. FloTrac-derived SVV values were recorded on a laptop computer using Edwards Lifesciences Multi-Data Logger (version 4.1; Edwards Lifesciences).

A pulse oximeter probe (LNCS; Masimo Corp.) was shielded against ambient light sources after being placed on the index finger of one hand and connected to a Masimo Radical-7 monitor with PVI software (V. PVI is a measure of the dynamic changes in perfusion index that occur during a complete respiratory cycle and has previously been described in detail.16 For the measurement of SpO2 via pulse oximetry, red and infrared light were used. A constant amount of light (direct current, DC) from the pulse oximeter is absorbed by the skin, other tissues and nonpulsatile blood, whereas a variable amount of light (alternating current, AC) is absorbed by the pulsating arterial inflow. For perfusion index calculation, the infrared pulsatile signal is indexed against the nonpulsatile infrared signal and expressed as a percentage [perfusion index = (AC/DC) × 100], reflecting the amplitude of the pulse oximeter waveform. The infrared signal is used because it is less affected by changes in arterial saturation than the red-light signal. Then, PVI calculation is accomplished by measuring changes in perfusion index over a time interval sufficient to include one or more complete respiratory cycles as PVI = [(perfusion indexmax − perfusion indexmin)/perfusion indexmax] × 100. The lower the number, the less variability there is in the perfusion index over a respiratory cycle. Pulse oximeter PVI values were recorded on a laptop computer using TrendCom MFC application software V (Masimo Corp.).

All haemodynamic variables were measured simultaneously after induction of anaesthesia, when CO had stabilized (T1). A second measurement was performed (T2) 1 min following volume replacement by infusion of 6% HES 130/0.4 (7 ml kg−1) at a rate of 1 ml kg min−1. Measurements were achieved in a haemodynamic steady state without any bolus application of vasoactive drugs. According to the available literature, patients were classified as responders to volume loading if the increase in SVI was at least 15%, or as nonresponders if the increase in SVI was less than 15%.11,12

Statistical analysis

A previous investigation, evaluating fluid responsiveness in mechanically ventilated patients,5 revealed that 19 patients were necessary to detect a relevant difference of 0.30 between the area under the receiver operating characteristic (ROC) curves for different variables of cardiac preload (SVV 0.90, two-tailed t-test, 5% α error rate, 80% power) with an allocation ratio being 0.33 and a correlation between the two diagnostic tests assumed to be 0.30 for the positive group and 0.20 for the negative group.

For statistical analysis, all volume variables were indexed to body surface area. Statistical analysis was performed using the SPSS 17.0 software (SPSS Inc., Chicago, Illinois, USA). After assessment of normal distribution by the Lilliefors modification of the Kolmogorov–Smirnov test and visual assessment of the histograms and the probability plots (Q–Q plots), Student's t-test was used to compare variables. Values of continuous data are presented as mean (SD) or as median (range) when appropriate; categorical variables are displayed as frequency distributions (n) and simple percentages (%). Haemodynamic variables before volume loading in responders and nonresponders were compared with a nonparametric Mann–Whitney test.

Linear regression analysis was performed between the baseline (T1) values of SVV, PVI and CVP and the percentage value of changes in SVI (ΔSVI). Prediction of fluid responsiveness based on SVV, PVI and CVP was tested by calculating and comparing the area under curve (area under ROC curve) for an SVI increase greater than 15%. Threshold values were determined by considering values that yielded the greatest sensitivity and specificity.21 The precision of the ROC analysis for area under curve, sensitivity and specificity, as well as for the positive and negative predictive values are reported as 95% confidence intervals (CIs). A P value less than 0.05 was regarded as significant.


Demographic data, perioperative risk factors and the performed surgical procedure for all patients are presented in Table 1.

Table 1
Table 1:
Patients' details

All haemodynamic variables changed significantly following volume replacement as displayed in Table 2. The volume-induced increase in SVI was at least 15% (range 17.6–57.1%) in 15 patients (responders) and less than 15% in five patients (nonresponders).

Table 2
Table 2:
Haemodynamic variables at sample points T1 and T2

Before volume loading, SVV and PVI were significantly higher in responders to volume than in nonresponders. After volume loading (mean 531 ± 132 ml), SVV and PVI presented significant changes in responders and nonresponders (Fig. 1 and Table 2).

Fig. 1
Fig. 1

Baseline SVV correlated significantly with ΔSVI (r = 0.80; P < 0.001) as did baseline PVI (r = 0.61; P < 0.004), whereas baseline values of CVP showed no correlation with ΔSVI (r = 0.18; P = 0.45) as displayed in Fig. 2.

Fig. 2
Fig. 2

The areas under the ROC curves, showing the ability to differentiate between responders and nonresponders, are shown in Fig. 3 and Table 3. There was no significant difference between the area under the ROC curve for SVV (0.993) and PVI (0.973) with respect to identifying a ΔSVI greater than 15%. The best threshold values to predict fluid responsiveness were more than 11% for SVV (sensitivity 100% and specificity 80%) and more than 9.5% for PVI (sensitivity 93% and specificity 100%). Based on these threshold values, positive and negative predictive values were 93% (70–99%) and 100% (40–100%) for SVV and 100% (77–100%) and 83% (36–99%) for PVI, respectively. The area under the curve for CVP (0.553) was significantly lower than the area under the curves for SVV and PVI (Table 3). The optimal threshold value was 10.5 mmHg for CVP, yielding a sensitivity of 66% and a specificity of 40%, and positive and negative predictive values of 77% (46–95%) and 28% (4–71%), respectively. Calculating the misclassification rates for all tested variables revealed 5% for SVV, 5% for PVI and 53% for CVP, respectively.

Fig. 3
Fig. 3
Table 3
Table 3:
Area under the ROC curves and cut-off values


The results of the present study suggest that arterial pressure-based SVV by the FloTrac/Vigileo system and noninvasively assessed PVI using Masimo plethysmographic waveform analysis have the potential to serve as valid indicators of fluid responsiveness in mechanically ventilated major surgical patients. In contrast, CVP as a static variable of cardiac preload showed no correlation with volume-induced changes in stroke volume and thus represents a poor variable for predicting fluid responsiveness.

Several studies have shown that dynamic variables of cardiac preload such as SVV and PPV are valuable indicators of fluid responsiveness during mechanical ventilation.3–5 However, most dynamic variables assessing fluid responsiveness are invasive, technically challenging or require additional catheters. This issue is of increasing interest because goal-directed intraoperative volume optimization using PPV as the haemodynamic endpoint has recently been reported to reduce perioperative morbidity and total cost of treatment in high-risk surgical patients.6 However, before investigating the effect of SVV-guided or PVI-guided intraoperative volume optimization, every new dynamic variable has to be validated to predict fluid responsiveness.

Over recent years, there has been a strong tendency in volumetric monitoring towards continuous and less invasive assessment of preload variables to facilitate measurement procedures as well as to minimize the patient's risk. Promising results for FloTrac-derived SVV as a dynamic variable of cardiac preload were recently demonstrated in patients with circulatory failure after liver transplantation11 and in preoperative patients scheduled for coronary artery bypass grafting.12,13 SVV threshold values of about 10%, reported in these studies, representing the trigger value at which a volume challenge is expected to increase stroke volume, are consistent with those of our investigation and accentuate the reliability and accuracy of this variable under different clinical conditions. The latest algorithm improvements of the response time to changes in vascular tone obviously enhanced the accuracy of SVV assessment using the FloTrac/Vigileo system as compared with the original version of the software (V 1.01 and V 1.03), with which FloTrac-derived SVV still failed to predict fluid responsiveness.22

Respiratory variations in the POP waveform amplitude (ΔPOP) have been recently shown to predict fluid responsiveness in mechanically ventilated patients.14 However, up to now plethysmographic waveform analysis required specific tools and software, which were not yet widely available and therefore could not be used at the bedside. The recently introduced Masimo Radical-7 monitor provides an automatic, continuous and noninvasive analysis of ΔPOP by calculating PVI, detecting the maximal and minimal perfusion index value over at least one respiratory cycle. Cannesson et al.16 primarily demonstrated that PVI can predict the cardiocirculatory response to volume loading in a haemodynamically stable setting of preoperative cardiac surgery. Although the areas under the ROC curves in this investigation are comparable to the results of the present study (0.927 vs. 0.973), threshold values to discriminate between responders and nonresponders to volume loading are different (>14 vs. >9.5%). As already shown for many variables of fluid responsiveness, threshold values can vary between studies and settings. Landsverk et al.23 recently described a large intraindividual and interindividual variability in ΔPOP compared with PPV in a heterogeneous group of ICU patients, which was confirmed by our findings. Although PVI seems to be able to identify responders to a volume challenge, threshold values of PVI to predict fluid responsiveness should be interpreted with caution and intraindividual variation of this variable should be documented.23 Differences between the two devices concerning their absolute and cut-off values to discriminate between volume-loading responders and nonresponders can additionally be explained by the specific differences in measurement techniques. Therefore, we waived performing a Bland and Altman analysis between the pooled data of SVV and PVI, as these variables are indices of two different methods to predict fluid responsiveness, the values of which are not expected to be the same.

In contrast to the dynamic variables of cardiac preload, CVP failed to predict fluid responsiveness in this investigation, confirming the results of previous studies regarding this topic.1,2 CVP depends on a variety of factors such as the vascular volume, the vascular tone, the myocardial function and the intrathoracic pressure. Additionally, CVP has been found to be of little help in assessing circulating blood volume, did not adequately reflect preload status and is, therefore, unsuitable for predicting ventricular response to fluid loading.1,2,24 This fact emphasizes that, despite their limitations, any dynamic variable of cardiac preload provides better performance in predicting fluid responsiveness than static preload variables.25

A limitation of the study is the small number of patients enrolled in the trial. Although the study seems to be adequately powered to answer its primary question, as post-hoc analysis of the data revealed a power of 0.99 (two-tailed t-test, 5% α error rate), the wide 95% CIs for specificity argue for a considerable uncertainty in the estimates. Moreover, despite misclassification rates of merely 5% for SVV and PVI, the small number of nonresponders in the study might be responsible for inappropriately high area under ROC curves with any derived cut-off values being questionable.

According to Lima and Bakker,26 the perfusion index reflects vascular reactivity that possibly affects the pulsatile absorption component. As PVI is a measure of the dynamic changes in perfusion index, patients should be studied under stable clinical conditions to avoid changes in vascular tone, for example through nociceptive input or the presence of spontaneous breathing activity, which limits the clinical relevance of the presented results. Additional factors that have to be considered when interpreting the plethysmographic waveform variations are the uncertain role of the venous element of the signal, its dependency on the site of measurement and obliteration of the plethysmographic signal during severe vasoconstriction.25

The possibility that measuring SVI and SVV by the FloTrac/Vigileo system and mathematical coupling between variables could be feasible should be taken into account. However, SVV is not based on absolute stroke volume, but is a function of their relative changes over the respiratory cycle. Therefore, FloTrac-derived SVV should be able to accurately reflect fluid responsiveness, even if the absolute value of stroke volume calculation differs from the clinical standard, which uses a PAC to perform the intermittent thermodilution technique.12 Performing clinical studies in a setting of elective major surgery is associated with several limitations, for example the restricted use of a PAC, which is not indicated in this group of patients.27

Although arterial pressure-derived SVV revealed the best correlation with volume-induced changes in stroke volume, the results of our study suggest that both less invasive assessable variables, SVV and PVI, have the potential to serve as useful indicators of fluid responsiveness in mechanically ventilated major surgical patients. Despite several limitations, PVI seems to provide a simple alternative for accurate, noninvasive, and continuous preload monitoring. Further studies are required to demonstrate the usefulness of both automatically calculated variables to guide volume optimization in critically ill patients to improve outcome.


C.W. received speaking fees from Edwards Lifesciences. The authors declare that there are no further competing interests.

Departmental funding supported this study financially: Department of Anesthesiology, University Hospital, Regensburg, Germany.


1 Marik PE, Baram M, Vahid B. Does central venous pressure predict fluid responsiveness? A systematic review of the literature and the tale of seven mares. Chest 2008; 134:172–178.
2 Kumar A, Anel R, Bunnell E, et al. Pulmonary artery occlusion pressure and central venous pressure fail to predict ventricular filling volume, cardiac performance, or the response to volume infusion in normal subjects. Crit Care Med 2004; 32:691–699.
3 Hofer CK, Müller SM, Furrer L, et al. Stroke volume and pulse pressure variation for prediction of fluid responsiveness in patients undergoing off-pump coronary artery bypass grafting. Chest 2005; 128:848–854.
4 Michard F, Teboul JL. Using heart-lung interactions to assess fluid responsiveness during mechanical ventilation. Crit Care 2000; 4:282–289.
5 Wiesenack C, Fiegl C, Keyser A, et al. Assessment of fluid responsiveness in mechanically ventilated cardiac surgical patients. Eur J Anaesthesiol 2005; 22:658–665.
6 Lopes MR, Oliveira MA, Pereira VO, et al. Goal-directed fluid management based on pulse pressure variation monitoring during high-risk surgery: a pilot randomized controlled trial. Crit Care 2007; 11:R100.
7 Perel A. The value of functional hemodynamic parameters in hemodynamic monitoring of ventilated patients. Anaesthesist 2003; 52:1003–1004.
8 Mayer J, Boldt J, Beschmann R, et al. Uncalibrated arterial pressure waveform analysis for less-invasive cardiac output determination in obese patients undergoing cardiac surgery. Br J Anaesth 2009; 103:185–190.
9 Mayer J, Boldt J, Wolf MW, et al. Cardiac output derived from arterial pressure waveform analysis in patients undergoing cardiac surgery: validity of a second generation device. Anesth Analg 2008; 106:867–872.
10 Prasser C, Trabold B, Schwab A, et al. Evaluation of an improved algorithm for arterial pressure-based cardiac output assessment without external calibration. Intensive Care Med 2007; 33:2223–2225.
11 Biais M, Nouette-Gaulain K, Cottenceau V, et al. Uncalibrated pulse contour-derived stroke volume variation predicts fluid responsiveness in mechanically ventilated patients undergoing liver transplantation. Br J Anaesth 2008; 101:761–768.
12 Cannesson M, Musard H, Desebbe O, et al. The ability of stroke volume variations obtained with Vigileo/FloTrac system to monitor fluid responsiveness in mechanically ventilated patients. Anesth Analg 2009; 108:513–517.
13 Hofer CK, Senn A, Weibel L, Zollinger A. Assessment of stroke volume variation for prediction of fluid responsiveness using the modified FloTrac and PiCCOplus system. Crit Care 2008; 12:R82.
14 Desebbe O, Cannesson M. Using ventilation-induced plethysmographic variations to optimize patient fluid status. Curr Opin Anaesthesiol 2008; 21:772–778.
15 Cannesson M, Delannoy B, Morand A, et al. Does the pleth variability index indicate the respiratory-induced variation in the plethysmogram and arterial pressure waveforms? Anesth Analg 2008; 106:1189–1194.
16 Cannesson M, Desebbe O, Rosamel P, et al. Pleth variability index to monitor the respiratory variations in the pulse oximeter plethysmographic waveform amplitude and predict fluid responsiveness in the operating theatre. Br J Anaesth 2008; 101:200–206.
17 Keller G, Cassar E, Desebbe O, et al. Ability of pleth variability index to detect hemodynamic changes induced by passive leg raising in spontaneously breathing volunteers. Crit Care 2008; 12:R37.
18 Manecke GR. Edwards FloTrac sensor and Vigileo monitor: easy, accurate, reliable cardiac output assessment using the arterial pulse wave. Expert Rev Med Devices 2005; 2:523–527.
19 Boulain T, Achard J-M, Teboul J-L, et al. Changes in BP induced by passive leg raising predict response to fluid loading in critically ill patients. Chest 2002; 121:1245–1252.
20 Langewouters GJ, Wesseling KH, Goedhard WJ. The pressure dependent dynamic elasticity of 35 thoracic and 16 abdominal human aortas in vitro described by a five component model. J Biomech 1985; 18:613–620.
21 Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148:839–843.
22 de Waal EE, Rex S, Kruitwagen CL, et al. Stroke volume variation obtained with FloTrac/Vigileo fails to predict fluid responsiveness in coronary artery bypass graft patients. Br J Anaesth 2008; 100:725–726.
23 Landsverk SA, Hoiseth LO, Kvandal P, et al. Poor agreement between respiratory variations in pulse oximetry photoplethysmographic waveform amplitude and pulse pressure in intensive care unit patients. Anesthesiology 2008; 109:849–855.
24 Wiesenack C, Prasser C, Keyl C, Rödīg G. Assessment of intrathoracic blood volume as an indicator of cardiac preload: single transpulmonary thermodilution technique versus assessment of pressure preload parameters derived from a pulmonary artery catheter. J Cardiothorac Vasc Anesth 2001; 15:584–588.
25 Perel A. Automated assessment of fluid responsiveness in mechanically ventilated patients. Anesth Analg 2008; 106:1031–1033.
26 Lima A, Bakker J. Noninvasive monitoring of peripheral perfusion. Intensive Care Med 2005; 31:1316–1326.
27 de Waal EE, Wappler F, Buhre WF. Cardiac output monitoring. Curr Opin Anaesthesiol 2009; 22:71–77.

fluid responsiveness; major abdominal surgery; pleth variability index; pulse oximeter plethysmography; stroke volume variation

© 2010 European Society of Anaesthesiology