This prospective observational study was approved by the Institutional Ethics Committee of Hitachi General Hospital, Japan (Approval No. 2020-48) and registered in the Clinical Trials Registry (ref: UMIN000044953). Written informed consent was obtained from the patients. Patients aged over 65 years who underwent TAVR under general anesthesia from September 2020 to October 2021 were included. Patients with a diagnosis of peripheral artery disease, Raynaud’s symptoms, emergency surgery and those who did not consent to the study were excluded. We performed standard monitoring during operation, including a 5-lead electrocardiogram, pulse oximeter and non-invasive intermittent BP measurement with the upper arm. Anesthesia was induced with propofol 1–2 mg.kg− 1 and fentanyl 0.05–0.1 mg.kg− 1. Tracheal intubation was facilitated by muscle relaxants. Ventilation was performed with a tidal volume of 7–8 ml.kg− 1 at ideal body weight and positive pressure ventilation of 5–10 cmH2O. Inhaled oxygen fractions and respiratory rate were adjusted to maintain peripheral oxygen saturation above 96% and end-expiratory partial pressure of carbon dioxide between 35 and 45 mmHg. After induction of anesthesia, the IAP was measured at the radial artery through a catheter (Terumo arterial catheter, 22-gauge, 23 mm length; Terumo, Shibuya, Tokyo, Japan) and the FloTrac TM (Edwards Lifesciences) pressure transducer connected to a module (Life Scope TR, Nihon Kohden Co, Sinjuku, Tokyo, Japan) for direct BP measurement. The transducer was placed at the level of the right atrium. The IAP waveform was visually assessed by the attending anesthetist to ensure that there was no dumping. For non-invasive BP monitoring, the ClearSight finger-cuff was placed on the index or middle finger, ipsilateral to the IAP monitoring, with the correct size as recommended by the manufacturer. All patients had their IAP monitored using FloTrac with Vigileo™ (Edwards Lifesciences) platform and non-invasive finger-cuff BP monitoring was done using ClearSight with HemoSphere™ (Edwards Lifesciences) platform recorded at the same time intraoperatively. All haemodynamic data were automatically recorded using information management systems (PrimeGaia™, Nihon Kohden). After the prosthetic valve has been deployed, All patients were confirmed by transoesophageal echocardiography (TEE) to have no more than mild aortic regurgitation of the prosthetic valve and no paravalvular leakage. The ejection fraction (EF) was measured by the modified Simpson method and the aortic valve area index (AVAI) of the prosthetic aortic valve was also measured by TEE using the continuous equation, the formula is as follows;
AVAI (cm2.m− 2) = CSALVOT (cm2) × VLVOT (m.s− 1)/VAV (m.s− 1)/BSA (m2)
CSALVOT: cross-sectional area of left ventricular outflow tract
VLVOT: blood flow velocity in the left ventricular outflow tract
VAV: blood flow velocity in aortic valve
BSA: body surface area
After TEE measurement, minute by minute IAP and ClearSight arterial pressure (CSAP) measurement were recorded for 30 mins and cardiac index (CI) (CIIAP and CICSAP) values were calculated by pulse contour method, which is the concept that the area of each arterial waveform corresponds to the stroke volume [8].
The sample size was calculated to be more than 646 pair data, based on the assumption that the two BP pair data to be compared would show a correlation coefficient of at least 0.8. Considering the deviation from the inclusion criteria, we decided to collect data from 30 patients. To assess the concordance of hemodynamic variables measured by IAP (reference method) and CSAP (test method), Bland–Altman analysis of repeated measurements was performed to calculate bias, precision, and limits of agreement [9, 10]. The percentage error was calculated as described by Critchley and Critchley [11]. A four-quadrant plot analysis was performed to evaluate the trend-tracking ability of the CSAP per minute with reference to the IAP [8]. The trend-tracking ability can be judged by the concordance rate, which is considered good if more than 92% of all values are in the upper right and lower left of the quadrant [11, 12]. The value in the center of the analysis table is set as the exclusion zone, which can be understood as the measurement point where the value did not change in the one-minute trend. The exclusion zone was set at 5 mmHg for BP data comparison [13, 14]. The error grid analyses were performed to compare the systolic and mean BPs of IAP and CSAP. The error grid analysis for arterial pressure can be performed to compare the clinical accuracy of BP estimates from a non-invasive measurement device with BP obtained with reference direct arterial pressure, reported by Saugel et al [15, 16]. The error between the gold standard and the test method was classified into five different clinical zones (from A to E) to assess the risk of leading to wrong intraoperative decisions [15]:
-
No risk (no difference in clinical actions between the test and gold standard methods).
-
Low risk (the values assessed by the test method and the gold standard differ, but the difference will probably lead to benign or no treatment)
-
Moderate risk (the values assessed by the test method and the gold standard differ, and the differences would lead to unnecessary treatment with moderate results that are not life-threatening to the patient)
-
Significant risk (the values assessed using the test method and the gold standard differ, and the difference leads to unnecessary treatment with serious non-life-threatening consequences for the patient).
-
Dangerous risk (the values assessed using the test method and the gold standard differ, and the difference leads to unnecessary treatment with life-threatening consequences for the patient).
In the error grid analysis, the test method can be considered to have the same clinical accuracy as the gold standard if the observed values that fall into the no risk category, i.e., zone A are 90% or more, zone B and C are within 5%, zone D is within 4%, and zone E is within 2% [15]. Logistic regression analysis was performed to explore potential confounding factors that could lead to errors in the IAP and CSAP being classified into a non-zone A category of clinical risk in the error grid analysis. The previously reported covariates included in the multivariate analysis were age, consecutive phenylephrine administration, coronary artery disease, hypertension, low CI and abnormal SVRI [5, 6, 17–19]. The patient population in this study was elderly and skewed by base hypertension and coronary artery disease immediately after valve replacement. After adjustment for patient background, AVAI, CI, and SVRI were selected as covariates. Statistical tests were two-tailed, and p < 0.05 was statistically significant. For statistical analysis, we used Matlab (The MathWorks Inc, Natick, MA, USA) and Stata/BE for Mac (Version 17.0; StataCorp, College Station, TX, USA).