Comparison of noninvasive and minimally invasive pulse contour analysis to measure stroke volume during major surgery

Editor, Goal-directed haemodynamic therapy (GDHT) has been shown to improve postoperative outcomes in major surgery but its implementation is limited, largely because of the invasiveness of monitoring devices.1 When compared with invasive devices, a noninvasive cardiac output (CO) monitor using photoplethysmography (Clearsight; Edwards Lifesciences, Irvine, Clifornia, USA) has shown a low systematic bias for cardiac index (CD). Its precision in the majority of studies is clinically unacceptable.2–5 Most studies have involved the postoperative period, and none have studied accuracy to predict fluid responsiveness during surgery. We compared haemodynamic variables and fluid responsiveness of dynamic indices between Clearsight and a minimally invasive pulse contour device (proAQT; Pulsion Medical Systems, Munich, Germany) during elective noncardiac major surgery. The study protocol was approved by the local Ethics Committee on 16 February 2016 (CPP Ouest III, Poitiers, France) and registered on ClinicalTrials.gov (number NCT02717468). After informed consent was obtained, 30 patients undergoing major elective surgery and requiring continuous invasive arterial pressure and CO monitoring were enrolled. After induction of anaesthesia (propofol, sufentanil or remifentanil target-controlled infusion and atracurium) and tracheal intubation patients were ventilated using volume-controlled mechanical ventilation with tidal volume of 7 to 8 ml kg−1 of ideal body weight and positive expiratory pressure of 5 to 7 cmH2O. An arterial catheter was inserted into a radial artery and connected to a proAQT sensor, plugged into the Pulsioflex monitor (Pulsion Medical Systems). The Clearsight Finger Cuff was applied to the mid-phalanx of the middle or index finger ipsilateral to the radial artery catheter and connected to the wrist unit, plugged into the EV1000 monitor (Edwards Lifesciences). Before recording measurements, the proAQT system was auto-calibrated and Clearsight was zeroed. Stroke volume index (SVI), CI and stroke volume variation (SVV) were collected by Pulsioflex and EV1000 monitors and recorded continuously (one measurement every 20 s) in the anaesthesia recording system (Diane, Version 4.4.5; Bow Medical, Bow, UK). During surgery, fluid management and use of vasopressors were left to the discretion of the attending anaesthetist but fluid expansion was standardised (250 ml of crystalloid, over 10 min) and haemodynamic variables were recorded immediately before and at the end of each therapeutic intervention. Two-way statistical analysis was performed with a P value of less than 0.05 considered as statistically significant using R 3.2.2 (R foundation, Vienna, Austria). Bias and precision as SD were calculated between proAQT and Clearsight measurements with variance adjustment due to unequal multiple measurements per patient.6 After infusion of 250 ml, positive response to fluid expansion was defined as an increase in SVI given by proAQT at least 10% from baseline. Mean changes in variables before and after fluid expansion were compared with a paired t test. Agreement between the two devices to track SVI changes after fluid expansion was assessed with Cohen's kappa coefficient. From February to July 2016, 30 adults (23 men, median [IQR] age = 67 [58 to 72] years) were enrolled. Twenty-four patients had at least one bolus of norepinephrine (median [IQR] total dose = 331 [7 to 1390] μg). We analysed 6087 pairs of data including 62 fluid expansions of which 34% were positive. Bias and precision for SVI, CI and SVV between Clearsight and proAQT were 0.02 ± 9.9 ml m−2, −0.02 ± 0.68 l min−1 m−2 and −0.84 ± 6.22%, respectively. Agreement between the two devices to predict SVI variation after fluid expansion was strong (κ = 0.81). When fluid expansion led to a positive response, fluid challenge induced a similar increase in SVI and decrease in SVV between Clearsight and proAQT. There was no significant modification in CI with either device (Table 1).Table 1: Haemodynamic variable responses after a 250-ml fluid expansion (n=62)For absolute haemodynamic variable values, our study showed good accuracy for SVI and SVV. However, although our results showed acceptable bias for CI (−0.02 l min−1 m−2), the limits of agreement were wide-ranging (−1.36 to 1.34 l min−1 m−2), a finding consistent with previous studies. Indeed, in comparison with pulmonary arterial catheter or transpulmonary thermodilution, Nexfin – previous name of Clearsight – was unreliable as a means of estimating absolute values of CI with a limit of agreement higher than 1.0 l min−1 m−2.2,3 Previously, few studies had compared the accuracy of Clearsight for monitoring CI and SVI during surgery,4,5 and none of them had assessed its effectiveness to track changes in SVI during fluid expansion as recommended by guidelines. In our study, positive fluid expansion was correctly identified by Clearsight with strong concordance compared with proAQT (Cohen's kappa coefficient = 0.81) and with a bias close to zero (0.02 ml m−2) for the absolute value of SVI. Our results are inconsistent with those of Fischer et al.2 who after cardiac surgery, concluded that the CI, pulse pressure variation and SVV given by Nexfin failed to predict fluid responsiveness in comparison with transpulmonary thermodilution. Our study has some limitations. First, fluid expansion was initiated at the discretion of the attending anaesthetist. Only 34% of fluid expansions were positive, and one-third of patients had no fluid expansion. Although this result could appear underwhelming, it is consistent with previous studies on GDHT and reflects a real-life assessment of devices. Second, our study is based on a comparison with a minimally invasive pulse contour analysis device which is not the gold standard in stroke volume (SV) measurement. We made this choice because it appeared to us more important to assay Clearsight with moderate-to-major surgery patients (for whom noninvasive monitoring should be indicated) and because it would not be ethical to propose invasive monitoring such as pulmonary arterial catheter or transpulmonary thermodilution. Moreover, in clinical practice, the ability to track haemodynamic variable changes during fluid expansion is more important than measurement of their absolute values. Lastly, we chose to use the same arm for Clearsight and proAQT sensors, whereas the arterial catheter may alter blood flow to the finger arteries. To limit this issue, we systematically performed an Allen's test before the insertion of the arterial catheter. Although 48% of anaesthesiologists have declared that the monitoring available is too invasive1 – which limits its use in common practice – a noninvasive pulse contour device such Clearsight could facilitate further expansion of SV monitoring. Acknowledgements relating to this article Assistance with the letter: we would like to thank Jeffrey Arsham, Bertrand Debaene and Olivier Mimoz for their assistance with the manuscript. Financial support and sponsorship: the study was supported by University Hospital of Poitiers. Conflicts of interest: MB received travel expenses from Edwards Lifesciences and Pulsion.