Patients
This study obtained the approval of the Institutional Review Board (Comité d’Ethique pour la Recherche en Anesthésie-Réanimation, IRB-00010254-2018-054). The patients were informed about the inclusion of their anonymized health data in our database (Written informed consent was waived by the Institutional Review Board).
Inclusion criteria were as follows: patients older than 18 years scheduled for neurosurgery, equipped with radial arterial catheter and cardiac output monitor, surgery in the supine position and the absence of arrhythmia.
Perioperative management
Patients were monitored with non-invasive blood pressure, SpO2 and ECG. Total intravenous anesthesia was used by target-controlled infusion of remifentanil and propofol. Patients were mechanically ventilated using a volume-control mode with a tidal volume of 6–8 ml/kg of ideal body weight, respiratory rate was adjusted to maintain normocapnia, positive expiratory pressure was set between 3 and 6 cmH2O, FIO2 was adjusted to maintain pulse oximetry above 95% and inspiratory/expiratory ratio was 0.5
Hemodynamic monitoring
A radial arterial catheter inserted after the induction of anesthesia was connected to a specific transducer (ProAQT®, Pulsion Medical System) for SVI, SVV and PPV monitoring. The initial value of cardiac output was estimated with a proprietary algorithm performing an “auto-calibration”. Stroke volume was then determined by pulse contour analysis. PPV and SVV were continuously displayed on the Pulsioflex® monitor. SVI, CI, PPV and SVV measurements using ProAQT® are an average during the last 12 s and are updated every second. SVI, CI, PPV and SVV are displayed beat-to-beat as a “sliding average” of 12 s. When extracting data via a USB port to an Excel file, values were displayed every 12 s.
Study design
After the “auto-calibration”, data were continuously recorded every 12 s on the Pulsioflex® monitor, and extracted using USB port as an Microsoft® Excel file for statistical analysis. Recording started at least 20 min after the onset of induction of anesthesia. Data were recorded during hemodynamic stability (defined as a change in mean arterial pressure and heart rate less than 5%). In practice, we made these recordings during the placement of surgical drapes. Patients with hemodynamic instability requiring a decrease (or an increase) in anesthesia drug dosage, fluid infusion or administration of vasopressors, were excluded. Another exclusion criterion was any change in ventilatory setting by the physician in charge of the patient.
Statistical analysis
Data are expressed as mean (95% confidence interval) or median (5–95th percentiles) according to variable distribution. LSC is the minimum change that needs to be measured by a device to recognize a real change. LCS was calculated for cardiac index, stroke volume index, PPV and SVV as previously described [15,16,17]:
$${\text{Coefficient}}\;{\text{of}}\;{\text{variation}}\;\left( {\text{CV}} \right) = {\text{standard}}\;{\text{deviation}}/{\text{mean}}\;{\text{of}}\;{\text{measurements}}$$
$${\text{Coefficient}}\;{\text{error}}\left( {\text{CE}} \right) = {\text{CV}}/\surd n \, \left( {n = {\text{number}}\;{\text{of}}\;{\text{measurements}}\;{\text{per}}\;{\text{patient}}} \right)$$
Least significant changes (LSC) = The LSC can be described by the following equation:
$${\text{LSC}} = {\text{CE}} \times 1. 9 6\times \surd 2.$$
To analyze the LSC of pulse contour ,data were recorded every 12 s, so five measurements were analyzed every minute. LSC calculation included 2 measurements for 30 s, 3 measurements for 45 s, 5 measurements for 1 min, 10 measurements for 2 min, 15 measurements for 3 min, etc. Average LSC on 15 min was calculated as the mean of individual LSC. The same applies for LSC calculation at each time-point. For example, LSC at 30 s was calculated as the mean of individuals LSC calculated at 30 s. To compare the LSC averages at each minute, and to take into account repeated measurements, we performed an ANOVA for repeated measurements. If one of the means differed statistically from the others, we performed a Tukey test with Bonferroni correction to take into account the multiplicity of tests.
Because LSC was based on standard deviation calculation, distribution of CI, SVI, PPV or SVV had to follow a normal distribution. According to the central limit theorem, as the number (size) of the independent variable increases, the more likely to obtain a normal distribution of the sample. A minimal number of 30 is currently accepted to assume a normal distribution. Thus, the number of subjects should be ≥ 30. After taking into account uninterpretable data or exclusion, we considered that 50 subjects were needed for this study.
Statistical analysis was performed using Medcalc (software 16.4.3; Mariakerke, Belgium) and R Development Core Team ([2008]. R: A language and environment for statistical computing; R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL).