Self-calibrating pulse contour cardiac output: do validation studies really show its clinical reliability?

Semi-invasive measurement of cardiac output based on pulse contour: a review and analysis

PURPOSE

The aim of this review was to provide a meta-analysis of all five of the most popular systems for arterial pulse contour analysis compared with pulmonary artery thermodilution, the established reference method for measuring cardiac output (CO). The five investigated systems are FloTrac/Vigileo(®), PiCCO(®), LiDCO/PulseCO(®), PRAM/MostCare(®), and Modelflow.

SOURCE

In a comprehensive literature search through MEDLINE(®), Web of Knowledge (v.5.11), and Google Scholar, we identified prospective studies and reviews that compared the pulse contour approach with the reference method (n = 316). Data extracted from the 93 selected studies included range and mean cardiac output, bias, percentage error, software versions, and study population. We performed a pooled weighted analysis of their precision in determining CO in various patient groups and clinical settings.

PRINCIPAL FINDINGS

Results of the majority of studies indicate that the five investigated systems show acceptable accuracy during hemodynamically stable conditions. Forty-three studies provided adequate data for a pooled weighted analysis and resulted in a mean (SD) total pooled bias of -0.28 (1.25) L·min(-1), percentage error of 40%, and a correlation coefficient of r = 0.71. In hemodynamically unstable patients (n = 8), we found a higher percentage error (45%) and bias of -0.54 (1.64) L·min(-1).

CONCLUSION

During hemodynamic instability, CO measurement based on continuous arterial pulse contour analysis shows only limited agreement with intermittent bolus thermodilution. The calibrated systems seem to deliver more accurate measurements than the auto-calibrated or the non-calibrated systems. For reliable use of these semi-invasive systems, especially for critical therapeutic decisions during hemodynamic disorders, both a strategy for hemodynamic optimization and further technological improvements are necessary.

Accuracy and precision of calibrated arterial pulse contour analysis in patients with subarachnoid hemorrhage requiring high-dose vasopressor therapy: a prospective observational clinical trial

Introduction

Calibrated arterial pulse contour analysis has become an established method for the continuous monitoring of cardiac output (PCCO). However, data on its validity in hemodynamically instable patients beyond the setting of cardiac surgery are scarce. We performed the present study to assess the validity and precision of PCCO-measurements using the PiCCO™-device compared to transpulmonary thermodilution derived cardiac output (TPCO) as the reference technique in neurosurgical patients requiring high-dose vasopressor-therapy.

Methods

A total of 20 patients (16 females and 4 males) were included in this prospective observational clinical trial. All of them suffered from subarachnoid hemorrhage (Hunt&Hess grade I-V) due to rupture of a cerebral arterial aneurysm and underwent high-dose vasopressor therapy for the prevention/treatment of delayed cerebral ischemia (DCI). Simultaneous CO measurements by bolus TPCO and PCCO were obtained at baseline as well as 2 h, 6 h, 12 h, 24 h, 48 h and 72 h after inclusion.

Results

PCCO- and TPCO-measurements were obtained at baseline as well as 2 h, 6 h, 12 h, 24 h, 48 h and 72 h after inclusion. Patients received vasoactive support with (mean ± standard deviation, SD) 0.57 ± 0.49 μg · kg^-1 · min^-1 norepinephrine resulting in a mean arterial pressure of 103 ± 13 mmHg and a systemic vascular resistance of 943 ± 248 dyn · s · cm^-5. 136 CO-data pairs were analyzed. TPCO ranged from 5.2 to 14.3 l · min^-1 (mean ± SD 8.5 ± 2.0 l · min^-1) and PCCO ranged from 5.0 to 14.4 l · min^-1 (mean ± SD 8.6 ± 2.0 l · min^-1). Bias and limits of agreement (1.96 SD of the bias) were −0.03 ± 0.82 l · min^-1 and 1.62 l · min^-1, resulting in an overall percentage error of 18.8%. The precision of PCCO-measurements was 17.8%. Insufficient trending ability was indicated by concordance rates of 74% (exclusion zone of 15% (1.29 l · min^-1)) and 67% (without exclusion zone), as well as by polar plot analysis.

Conclusions

In neurosurgical patients requiring extensive vasoactive support, CO values obtained by calibrated PCCO showed clinically and statistically acceptable agreement with TPCO-measurements, but the results from concordance and polar plot analysis indicate an unreliable trending ability.

Inferior vena cava variation compared to pulse contour analysis as predictors of fluid responsiveness: a prospective cohort study

BACKGROUND

Both occult hypoperfusion and volume overload are associated with increased morbidity and mortality in critically ill patients. Accurately predicting fluid responsiveness (FRes) allows for optimization of cardiac performance while avoiding fluid overload and prolonged mechanical ventilation.

OBJECTIVE

To simultaneously assess the ability to predict FRes using the stroke volume variation (SVV) obtained with the Vigileo/Flotrac monitor and inferior vena cava respiratory variation (ΔIVC) measured by standard echocardiography ([ECHO) during mechanical ventilation.

METHODS

We included medical intensive care unit (ICU) patients undergoing mechanical ventilation that required vasopressors, had worsening organ function, and that were well adapted to the ventilator. We excluded patients requiring escalating doses of vasopressors, hemodialysis, with ascites and patients with atrial fibrillation or a heart rate >120/min. Stroke volume index (SVI) and SVV were obtained from the Vigileo monitor whereas ΔIVC was obtained with ECHO (M-mode). Doppler ECHO was used to measure SVI and used to determine FRes (defined by SVI increase ≥ 10%). A data set was obtained before and 30 minutes after a 10-minute fluid challenge (FC) with 500 mL of saline.

RESULTS

In all, 25 patients were prospectively enrolled over an 8-month period. A total of 12 patients had acute respiratory distress syndrome (ARDS), 3 had a cardiac arrest, and 10 had sepsis. The patients' mean age was 61.36 years (±13.7), study enrollment since ICU admission was 3.4 days (±3.39), the Sequential Organ Failure Assessment (SOFA) score was 12.44 (±2.59), and the tidal volume 8.6 mL/kg (±1.68). Of the 25 patients, 8 (32%) were FRes. The correlation coefficient between the baseline ΔIVC and percentage increase in SVI (by ECHO) after an FC was R(2) = .51 with a receiver operating characteristic (ROC) curve of 0.81 while that for the baseline SVV by Vigileo was R(2) = .12 with an ROC curve of 0.57. The mean SVI bias between ECHO and Vigileo was -2 mL/m(2), the precision was -18 to 14 and the mean error was 46%.

CONCLUSIONS

ECHO assessment of the IVC variation during mechanical ventilation may prove to be a useful technique to predict FRes and guide fluid resuscitation in the ICU. The SVV obtained with the Vigileo monitor failed to predict FRes likely due to lack of calibration and the use of a complex algorithm that may be unreliable in patients with sepsis.

Performance of cardiac output measurement derived from arterial pressure waveform analysis in patients requiring high-dose vasopressor therapy

BACKGROUND

Arterial pressure waveform analysis of cardiac output (APCO) without external calibration (FloTrac/Vigileo™) is critically dependent upon computation of vascular tone that has necessitated several refinements of the underlying software algorithms. We hypothesized that changes in vascular tone induced by high-dose vasopressor therapy affect the accuracy of APCO measurements independently of the FloTrac software version.

METHODS

In this prospective observational study, we assessed the validity of uncalibrated APCO measurements compared with transpulmonary thermodilution cardiac output (TPCO) measurements in 24 patients undergoing vasopressor therapy for the treatment of cerebral vasospasm after subarachnoid haemorrhage.

RESULTS

Patients received vasoactive support with [mean (sd)] 0.53 (0.46) µg kg(-1) min(-1) norepinephrine resulting in mean arterial pressure of 104 (14) mm Hg and mean systemic vascular resistance of 943 (248) dyn s(-1) cm(-5). Cardiac output (CO) data pairs (158) were obtained simultaneously by APCO and TPCO measurements. TPCO ranged from 5.2 to 14.3 litre min(-1), and APCO from 4.1 to 13.7 litre min(-1). Bias and limits of agreement were 0.9 and 2.5 litre min(-1), resulting in an overall percentage error of 29.6% for 68 data pairs analysed with the second-generation FloTrac(®) software and 27.9% for 90 data pairs analysed with the third-generation software. Precision of the reference technique was 2.6%, while APCO measurements yielded a precision of 29.5% and 27.9% for the second- and the third-generation software, respectively. For both software versions, bias (TPCO-APCO) correlated inversely with systemic vascular resistance.

CONCLUSIONS

In neurosurgical patients requiring high-dose vasopressor support, precision of uncalibrated CO measurements depended on systemic vascular resistance. Introduction of the third software algorithm did not improve the insufficient precision (>20%) for APCO measurements observed with the second software version.