System for measuring the mean arterial pressure

Abstract

The present invention relates in particular to the field of anesthesia and to a method for real-time evaluation of the mean arterial pressure of a patient from plethysmography measurements. It also relates to a method for treating a patient comprising by continuously evaluating the mean arterial pressure of the patient, based on values continuously calculated by plethysmography.

Claims

1. An ex vivo method of alerting a hypotensive state of a patient comprising the implementation of an ex vivo method for continuously evaluating the mean arterial pressure of a patient, based on values of a parameter Vp continuously calculated by plethysmography, the method comprising attaching a plethysmograph to the patient and: I. calculating a calibration value Calib from (a) the value of the mean arterial pressure measured at a time t0, and (b) the value Vp0 linked to said parameter, measured in the patient at the time t0, wherein Calib=(value of the mean arterial pressure measured at a time t0)/Vp0 and II. calculating an estimated value MAPest of the patient's arterial pressure at a time t after t0 by the formula MAPest=Calib×Vpt, wherein Vpt is a value of a measurement of dicrotic wave height obtained at the time t, or a value averaged from a plurality of measurements obtained over a predetermined period of time, and emitting a signal when the value MAPest is below a predetermined threshold.

2. The method of claim 1, wherein the plethysmograph is attached to a finger or an earlobe of the patient and the value Vpt is obtained by photoplethysmography.

3. The method of claim 1, further comprising calculating the value of the perfusion index, the inverse of the perfusion index or the logarithm of the inverse of the perfusion index+1.

4. The method of claim 1, wherein the value Vpt is a value averaged from a plurality of measured values over a predetermined period of time.

5. The method of claim 1, further comprising recalculating the value Calib by measuring a new arterial pressure value and measuring a new value of the parameter Vp, and subsequently using the recalculated value Calib.

6. The method of claim 1, which is carried out during an entire period of a general anesthesia of the patient.

7. An ex vivo method of alerting a hypotensive state of a patient comprising the implementation of an ex vivo method for continuously evaluating the mean arterial pressure of a patient, based on values of a parameter Vp continuously calculated by plethysmography, the method comprising attaching a plethysmograph to the patient and: I. calculating a calibration value Calib from (a) the value of the mean arterial pressure measured at a time t0, and (b) the value Vp0 linked to said parameter, measured in the patient at the time t0, wherein Calib=(value of the mean arterial pressure measured at a time t0)/Vp0 and II. calculating an estimated value MAPest of the patient's arterial pressure at a time t after t0 by the formula MAPest=Calib×Vpt, wherein Vpt is a value of a measurement of dicrotic wave height obtained at the time t, or a value averaged from a plurality of measurements obtained over a predetermined period of time, and III. treating the patient when the estimated mean arterial pressure (MAPest) is below a predetermined threshold.

8. The ex vivo method of claim 7 wherein a vasopressor is administered to the patient.

9. The method of claim 7, wherein the plethysmograph is attached to a finger or an earlobe of the patient and the value Vpt is obtained by photoplethysmography.

10. The method of claim 7, further comprising calculating the value of the perfusion index, the inverse of the perfusion index or the logarithm of the inverse of the perfusion index+1.

11. The method of claim 7, wherein the value Vpt is a value averaged from a plurality of measured values over a predetermined period of time.

12. The method of claim 7, further comprising recalculating the value Calib by measuring a new arterial pressure value and measuring a new value of the parameter Vp, and subsequently using the recalculated value Calib.

13. The method of claim 7, which is carried out during an entire period of a general anesthesia of the patient.

Description

BRIEF DESCRIPTION OF THE FIGURES

(1) FIG. 1: Example of a graph of a sphygmomanometer signal (obtained from https://www.infirmiers.com/etudiants-en-ifsi/cou rs/cours-card iologie-la-pression-arterielle-et-sa-mesure.html)

(2) FIG. 2: Principle of pulse oximetry. (1): variable light absorption related to the variation in arterial blood volume. (2) constant light absorption related to the non-pulsatile portion of arterial blood. (3): constant light absorption related to venous blood. (4): constant light absorption related to tissue, bone, etc. According to Feissel, Réanimation 16 (2007) 124-131.

(3) FIG. 3: Representation of the variables usable in the context of the invention, based on a pulse oximeter plot. T1: duration between the beginning and the maximum of the pulse wave; T2: duration between the beginning of the pulse wave and the dicrotic wave; T3: total duration of the pulse wave; Hd: dicrotic wave height.

(4) FIG. 4: Representations of different types of plethysmographic signals with dicrotic wave identification.

(5) FIG. 5: Flowchart representing the implementation of a process according to the invention.

(6) FIG. 6: Flowchart representing the implementation of another embodiment of a process according to the invention.

(7) FIG. 7: A. Representation, in a representative patient, of MAP measured by an invasive method (ARTm, large dotted line), MAP estimated by a method according to the invention on the basis of the dicrotic wave height (PlethoMAP, small dotted line), MAP measured by the cuff sphygmomanometer (NBPm, black dots) and the calibration factor (Calib, solid line). B. Representation, for the same patient and the same period, of the change in the systolic peak (solid line), the dicrotic wave (small dotted line) and the diastole (large dotted line) over time. A graphical representation of the superimposed pulse wave is also shown in this figure, solely to improve understanding of the three points (what each value corresponds to). Note: the time scales of this pulse wave representation and the change in the peaks over time are different and this pulse wave representation (usually lasting 1 sec or less) is only present for information purposes.

(8) FIG. 8: Flowchart representing the implementation of an embodiment of analysis of the plethysmographic signal according to a process according to the invention.

(9) FIG. 9: Definition of Dicpleth and PI. a=amplitude of the pulsatile component of the photoplethysmographic signal (height of the systolic peak, pulsatile part); b=height of the dicrotic notch (pulsatile part); c=amplitude of the stationary component of the photoplethysmographic signal.

(10) FIG. 10: Median values of ΔMAP and ΔDicpleth (A) and ΔMAP and ΔPI (B) during induction of anesthesia. ΔDicpleth: relative change in Dicpleth versus baseline; ΔPI: relative change in perfusion index versus baseline; ΔMAP: relative change in MAP versus baseline. The figure represents the change in ΔMAP and ΔDicpleth during the 16 min induction of anesthesia. The median MAP and the variation of the Dicpleth relative to the baseline are represented every minute (T “x”, at “x” min from T0) from the start of induction (T0).

(11) FIG. 11: Flowchart representing the implementation of a process according to the invention, using the measurement of the height of the dicrotic notch and the perfusion index.

EXAMPLES

(12) The examples below illustrate different aspects and modes of implementation of the invention. The embodiments described in the examples are an integral part of the invention.

Example 1. Determination of Parameters that can be Used for Continuous Arterial Pressure Measurement

(13) In cardiology, it is thus possible to measure the variation of the pulse wave non-invasively by plethysmography. This produces a curve (graph) representing the excess volume due to systolic expulsion.

(14) The perfusion index (PI) reflects the amount of blood flow measured locally and in part of the pulsating arterial flow, the systolic ejection volume. The perfusion index represents the area under the curve mentioned above. In the case of photoelectric plethysmography, a value SpO.sub.2 is also obtained, representing the arterial oxygen saturation.

(15) The curve represents the profile of the pulse wave and shows the dicrotic wave at the output of the heart at systole, (a second peak (optionally two peaks), a plateau or a break in the decay) in the organ.

(16) This plethysmographic signal can be used for continuous mean arterial pressure measurement.

(17) Thus, variations in peak size (of the dicrotic wave), area (value of the perfusion index) or timing between two events in the pulse wave profile are used. In particular, the decay of the dicrotic wave can actually be seen, i.e. the dicrotic wave can be detected with certainty in the plethysmographic signal.

(18) In order to determine these parameters, it is possible to analyze the pulse wave.

(19) Foot of the Pulse Wave

(20) First, the foot of the pulse wave can be detected. The foot of the pulse wave is characterized by a rapid rise in the signal, resulting in a peak of the second derivative of the signal. The analysis of this second derivative of the signal obtained by the plethysmograph makes it possible to obtain a signal (peak of the second derivative) at each rising edge of the pulse wave, and thus to detect the foot of the pulse wave, and thus the moment corresponding to the beginning of the signal.

(21) Systole (Pulse Wave Peak)

(22) From the foot of the wave, for each cycle, the maximum value of the cycle is sought. To do this, the signal can be split into several time windows and the maximum value in each time window sought. In this way the local maximum value corresponding to the maximum of the pulse wave can be determined. The result is as follows the value of the maximum (absorbance value given by the pulse oximeter corresponding to the peak of the systolic wave) the time at which this maximum is reached

(23) The time between the start of the pulse wave and the maximum of the pulse wave can therefore be calculated.

(24) Dicrotic Wave

(25) Once the peak of the pulse wave has been identified, the second derivative of the signal is analyzed over predetermined time windows (between 50 ms and 300 ms). The local maximum of the second derivative which is after the systolic peak is sought, in order to determine an area of interest in which the absolute minimum of the first derivative is searched for. The dicrotic point corresponding to the dicrotic wave corresponds to this point where the absolute minimum of the first derivative is reached. Its value (absorbance value given by the pulse oximeter), as well as the time between the foot of the pulse wave and this point, can then be measured.

(26) Exemplification of Practical Application of this Method

(27) Signal Collection

(28) The signal was collected in real time from a standard patient monitor capable of providing a photoplethysmography waveform as well as non-invasive arterial pressure by means of a pressure cuff. Connection to the monitor is usually via an RS232 serial port or a network connection via ethernet or WiFi. In addition to these two essential parameters, the perfusion index (PI) value was also used. The communication with the monitor can be bidirectional and allow, for example, a new non-invasive pressure tap to be requested on demand.

(29) Signal Analysis

(30) The signal processing is based on an online algorithm that measures the measured values and produces a beat-by-beat result in real time.

(31) Foot of the Pulse Wave

(32) The software uses a heartbeat detection algorithm that relies on the detection of the foot of the pulse wave. The foot of the pulse wave is characterized by a rapid rise in the signal, resulting in a peak of the second derivative of the signal.

(33) The detection of the foot of the pulse wave is based on the second derivative of the signal. This second derivative is weighted by the first derivative to focus only on the increasing part of the signal (i.e. the second derivative is given a zero value when the first derivative is not positive). The values obtained are squared and then float-integrated over a 240 ms centered window (averaged over values 120 ms before and 120 ms after the desired point). This integrated signal provides a powerful signal at each rising edge of the pulse wave.

(34) This signal is compared with a threshold value. This threshold value depends on the patient, the equipment used, the shape of the plethysmographic signal and the measurement noise. Since the measurement noise is not constant, the threshold is necessarily adaptive in real time. To calculate the threshold, the integral (calculated above) is used to which a floating mean with a window of 3 s (centered or not) is applied, the result being multiplied by 1.5. The threshold thus obtained can be used to define an area of interest where the integral exceeds the threshold. Within this area of interest, the peak of the second derivative defines the foot of the wave.

(35) Systole (Pulse Wave Peak)

(36) From the foot of the wave, for each cycle, the detection of the dicrotic wave is done in two steps. The first step consists in finding the systole and thus the maximum value of the cycle. To do this, the signal is split into 50 ms windows and the signal is advanced window by window as long as higher values are found. Once the highest value has been exceeded, the local maximum value (corresponding to the maximum value of the pulse wave, or systole value) is found.

(37) Dicrotic Wave

(38) Once the peak of the pulse wave has been identified, the next step is to analyze the second derivative of the signal as it progresses through 150 ms windows. In this way the local maximum of the second derivative, which is located after the systolic peak, is sought. Once this peak has been identified, it informs of an area of interest in which the dicrotic wave is located. From this point, the first derivative is analyzed, and the absolute minimum of the first derivative is searched for within an 8 ms window from the peak of the second derivative.

(39) The minimum of the first derivative close to the peak of the second derivative is thus obtained. This point is defined as the dicrotic point corresponding to the dicrotic wave and its value can be measured (total absorbance value given by the pulse oximeter).

Example 2. Calibration and Continuous Estimation of Mean Arterial Pressure (MAP)

(40) Calibration requires the value (Vp) of at least one of the following parameters dicrotic wave height value of the logarithm (natural or decimal) of the inverse of the PI perfusion index (ln (1/PI+1)). 1 is added to the inverse of the perfusion index to avoid taking the logarithm of a value less than 1 and obtaining a negative result value of the time between the foot of the pulse wave and the dicrotic wave and value of the time between the foot of the pulse wave and the maximum of the pulse wave ratio “dicrotic wave height/pulse wave height” and/or “dicrotic wave height/diastole wave height”. total pulse wave duration

(41) These values can be obtained beat-by-beat (i.e. for each pulse wave). FIG. 3 shows how these variables are measured.

(42) Calibration also requires the mean arterial pressure (MAP) value, which can be obtained, for example, with a non-invasive pressure cuff.

(43) A mean over several cycles (2, 3, 4, 5, 6, 8 or 10 cycles) of the value of the selected parameter can be used. This avoids disturbances such as respiratory pressure variability and irregular heart cycle. This mean is preferably the statistical median rather than the arithmetic mean.

(44) A calibration factor Calib is estimated when taking non-invasive arterial pressure and is obtained by Calib=MAP/Vp. It is understood that the value Calib depends on the parameter chosen and that the value Calib obtained if the dicrotic wave value is chosen will be different from the value Calib if the logarithm of (the inverse of the perfusion index (PI)+1) is chosen.

(45) Once the value Calib is obtained, the mean arterial pressure is estimated beat-by-beat using the photoplethysmographic signal as the sole data source.

(46) The estimated MAP (MAPest) at the time t is calculated by the formula
MAPest=Calib×VPt, where Vpt is the value (optionally averaged) of the chosen parameter.

Application Example (the Parameter being the Dicrotic Wave)

(47) Calibration requires the values of the beat-to-beat dicrotic wave as well as intermittent values of mean arterial pressure, for example by a non-invasive pressure cuff. The dicrotic pressure value obtained by the dicrotic wave is averaged over several cycles.

(48) The number of cycles over which the value is averaged is adjustable (for example 5 cycles). This prevents disturbances such as respiratory pressure variability and irregularities in the cardiac cycle. Averaging is done using the statistical median rather than the arithmetic mean to be more robust in the presence of noise. A calibration factor is estimated when taking non-invasive arterial pressure and is based on the current averaged dicrotic value (Pdic) and the measured mean arterial pressure (MAP). The calibration factor is obtained by Calib=MAP/Pdic.

(49) Continuous Estimation of Mean Arterial Pressure (MAP)

(50) Once calibration is performed, the mean arterial pressure is estimated beat-by-beat using the absorption measured by photoplethysmography as the sole signal source. The estimated MAP (MAPest) is calculated by MAPest=Calib.Math.Pdic, where Pdic is the averaged value of the dicrotic wave value as described above.

Another Example (Use of the Perfusion Index), Especially as a Signal Quality

(51) The PI value serves as an indicator of signal quality. Indeed, a PI value below 0.1% indicates a poor photoplethysmographic signal and can indicate to the user poor estimation quality and the more frequent need for calibration. Signal quality can usually be improved in these cases by repositioning the sensor correctly on the patient.

(52) Incorporation into the Estimate

(53) The PI gives information on the hemodynamic status in the mean arterial pressure in the same way as the dicrotic wave and in a complementary way. Indeed, the PI generally evolves in the opposite direction to the mean arterial pressure.

(54) A measure referred to as mPI (for modified PI) is used and is calculated as follows: mPI=10×ln(1/PI+1). The mPI thus obtained varies in the same direction as the mean arterial pressure and the behavior is linearized with respect to the exponential behavior of PI.

(55) A calibration Calib is performed with the measured mean arterial pressure and the mPI at the time 0 and the mean arterial pressure at the time t is calculated using MAPest=Calib×mPI(t).

(56) Using Multiple Parameters

(57) Several parameters can be used (dicrotic wave height, mPI, times indicated above).

(58) One can regularly calculate a Calib for each parameter calculate the MAPest for each of the parameters define the MAPest by statistical mean (Kalman filter) by weighting these values MAPest and using the value calculated at the previous time

Example 3. Exemplification in Real Conditions

(59) These results were obtained on the basis of the dicrotic wave height, similar results can be obtained with the other parameters.

(60) A study was conducted in the neurosurgical operating room or during an interventional neuroradiology procedure. Patients received the usual basic management for this type of intervention including:

(61) Monitoring Non-invasive hemodynamic arterial pressure monitoring by oscillometry and continuous ECG monitoring Continuous monitoring of arterial oxygen saturation (SpO.sub.2) by photoplethysmography, as well as monitoring of expired CO2. Bispectral index (BIS) monitoring of the depth of anesthesia Induction and maintenance of general anesthesia using target controlled infusion (TCI) including propofol and remifentanil, and curarization with atracurium besylate prior to orotracheal intubation. All monitors were connected to a Philips monitor. Hypotension was defined as a decrease in mean arterial pressure (MAP) of at least 20% compared with basal MAP. In the event of hypotension, the anesthetist in charge of the patient was free to lighten the anesthesia, to administer a vascular filling or a vasoconstrictor (ephedrine 9 mg, phenylephrine 50 mcg or noradrenaline 10 mcg).

(62) Experimental Protocol

(63) Phase 1: Pre-Oxygenation (Baseline)—Anesthetic Induction. Pre-oxygenation for 2 min: It is during this phase and before any injection that the basic values of all the parameters are recorded (mean of 2 values, baseline). Remifentanil at 5 ng/mL target concentration for 1 min. Propofol 5 μg/mL target concentration. After the BIS has dropped below 50 and checking for absence of ciliary reflex and the patient is manually ventilatable: curarization with 0.5 mg/kg of tracrium. Waiting time of 3 min with manual ventilation.

(64) Phase 2: Laryngoscopy-Intubation-Manual Ventilation. Direct laryngoscopy. Orotracheal intubation. Manual ventilation and probe attachment.

(65) Phase 3: Mechanical Ventilation-Anesthesia Maintenance. The patient is connected to the ventilator and mechanical ventilation started. Remifentanil and propofol targets decreased to 3.5 ng/mL and 4 μg/mL, respectively. Continuation of the collection for 3 min.

(66) Phase 4: Possible Hypotension Correction by Vasoconstrictor. Hypotensive episode treated with vasoconstrictor. Continuation of the collection one minute after the vasoconstrictor takes effect. End of collection.

(67) During phases 1, 2 and the beginning of 3 the cuff pressure was taken every minute for an estimated mean duration of 15 minutes. The anesthetist in charge of the patient was free to deviate from the initial protocol at any time if he considered that the clinical situation required it.

(68) Data collection was performed using Extrend data acquisition software (Ixellence), collecting signals at a frequency of 125 Hz and all numerical values.

(69) A collection point of all the following parameters was carried out every minute: Dicrotic wave height: the calculation of the dicrotic wave height on the plethysmography signal. PI: the perfusion index was collected beat-by-beat.

(70) Results:

(71) 61 patients were included in the study (median age 55 years, 32.7% males/67.3% females).

(72) 54 out of 61 patients had at least one hypotensive episode. The incidence of hypotension, defined as a decrease in MAP>20%, in the population was 88.5%. The time spent with MAP<20% is on mean 5.2 min during induction, i.e. 44% of the time.

(73) Evolution of the Values During the Entire Induction.

(74) The mean duration of the entire induction phase was 12±4 min.

(75) MAP Change and Dicrotic Wave Height

(76) MAP variations and variations in dicrotic wave height were highly correlated in a linear fashion over the induction period (see FIG. 7, especially the stability of the value Calib (FIG. 7.A)).

(77) Analysis Based on Continuous Mean Arterial Pressure Measurement by Invasive Arterial Catheterization.

(78) The arterial catheter is a device that allows arterial access to measure arterial pressure, invasively and continuously, and to take arterial blood samples.

(79) Bloody arterial pressure, or invasive arterial pressure, is an invasive technique for monitoring intravascular arterial pressure through an arterial catheter.

(80) A continuous measurement of arterial pressure was carried out by means of an arterial catheter and by the method according to the invention (calculation via the dicrotic wave height measured by plethysmography).

(81) A perfect correlation was observed between changes in MAP measured by arterial catheter and method after the use of vasopressor drugs. The correlation was r=0.88 with 96% agreement between the variations obtained by the technique and the actual MAP measurement (FIG. 7.A).

(82) In FIG. 7.A, a significant pressure variation can be seen at the end, which is not detected by the cuff (due to the time between two pressure measurements), but is detected by the PlethoMAP signal, thus stressing the informative nature of the method using the dicrotic wave.

(83) In FIG. 7.B, it can be seen that only the measurement of the dicrotic wave height allows a MAP value to be obtained, and that the change in the other two parameters (systolic or diastolic value) is not sufficiently informative. It can also be seen (T=1.15 h) that the drop in dicrotic wave height is indicative of the actual drop in MAP (FIG. 7.A), whereas systolic and diastolic pressures do not vary. Therefore, the combination of these two pressures would not have detected the decrease in mean arterial pressure and would not have allowed the anesthesiologist to take any corrective action.

Example 4. Other Parameters

(84) A study was conducted on patients in accordance with the applicable rules. The patients were over 18 years of age and underwent elective neuroradiological interventions after informed consent. The exclusion criteria for the study were cardiac arrhythmia (i.e. atrial fibrillation) and pregnancy.

(85) Anesthesia Protocol

(86) Prior to induction of anesthesia, standard monitoring was initiated by electrocardiogram, a non-invasive measurement of brachial ABP (PHILIPS FRANCE, Suresnes, France) set to inflate every minute, and digital pulse oximetry (PHILIPS FRANCE, Suresnes, France) placed on the second finger on the contralateral side of the ABP cuff. The bispectral index (BIS™quatro sensor, Medtronic France, Boulogne-Billancourt, France) and neuromuscular blockage monitoring (TOF Watch®, ALSEVIA PHARMA, Paris, France) were also used to monitor anesthesia. All monitoring parameters were available on a PHILIPS Intellivue MP 60 monitor (PHILIPS FRANCE, Suresnes, France). Induction of anesthesia was performed with remifentanil and propofol with an initial dose of 5 ng/mL and 5 μg/mL, respectively, and adjusted to achieve a BIS between 40 and 60. After the BIS decreased below 60 and loss of consciousness, neuromuscular blocking was performed by intravenous injection of 0.5 mg/kg of atracurium. Patients were then mechanically ventilated by tracheal intubation by direct laryngoscopy (end-tidal volume=6 mL/kg ideal body weight, positive end-tidal expiratory pressure=5 cmH.sub.2O, respiration rate and oxygen fraction to achieve end-tidal CO.sub.2=4.7 kPa and O.sub.2 saturation >95%).

(87) Arterial pressure was measured every minute during induction and every 5 minutes after tracheal intubation and stabilization. The patient's anesthetist may at any time change the frequency of measurements and treat 10H episodes with fluid loading and/or vasopressors (phenylephrine and/or norepinephrine). After induction, some patients may also benefit from continuous invasive arterial pressure monitoring.

(88) Data Collection

(89) All parameters and monitoring curves displayed on the screen were recorded on a computer. Hemodynamic parameters (heart rate, systolic arterial pressure [SAP], mean arterial pressure [MAP] and diastolic arterial pressure [DAP]) and PPG parameters (Dicpleth, PI and SpO.sub.2) were then sampled retrospectively every minute during induction. The induction period was arbitrarily set from pre-oxygenation to 3 min after connection to mechanical ventilation. Baseline values were obtained by averaging the two measurements (one minute apart) prior to the injection of anesthetics during the pre-oxygenation period. “Pre-pressor” values were defined as the pre-bolus vasopressor measurements during IOH episodes. “Peak pressure” values were defined as the maximum effects of the vasopressor bolus when the highest MAP was reached. Consistent with most studies, IOH was defined as a decrease of more than 20% from baseline MAP.

(90) Dicpleth and PI Measurement

(91) Dicpleth was obtained a posteriori from PPG waveforms recorded by an operator blind to ABP values. Dicpleth was defined as the ratio of the height of the dicrotic notch (from the nadir point of the complex to the notch) to the height of the systolic peak (from the same nadir point of the complex to the image), measured at the end of expiratory time in patients on mechanical ventilation (mean of 3 consecutive complexes) (FIG. 9). The PI (perfusion index) was provided by the manufacturer and is calculated as the ratio of the pulsatile component of the PPG signal to the DC component (FIG. 9).

(92) ΔMAP, ΔDicpleth and ΔPI were calculated during the induction period as their relative percentage changes from their reference values. During the vasopressor boluses, variations in the respective parameters were calculated between the “pre-pressor” and “peak-pressor” measurements.

(93) Dicradial Measurement

(94) In patients with invasive monitoring during maintenance of anesthesia, Dicradial was also measured from the arterial pressure signal using the same methodology as Dicpleth. The last 3 heartbeats of the end of the expiratory period were used to calculate Dicradial from the height of the dicrotic notch to the height of the systolic peak. In these patients, Dicpleth, Dicradial and their relative variations (ΔDicpleth and ΔDicradial) during vasoconstrictor administration were also analyzed.

(95) Statistical Analysis

(96) Values were expressed as median and interquartile range [25.sup.th and 75.sup.th percentiles]. Parameter changes were analyzed using the Wilcoxon ranking test. Percent agreement between the delta ΔMAP, and ΔDicpleth and ΔPI were calculated during the induction period. The areas under the curve (AUC) of the receiver operating characteristic (ROC) curve (with a 95% confidence interval) of ΔDicpleth and ΔPI to detect IOH episodes were estimated and optionally compared using the DeLong test. The Youden method was used to determine the optimal threshold values of ΔDicpleth and ΔPI for detecting IOH episodes. The combination ROC curve of ΔDicpleth and ΔPI was constructed using the logistic model. Correlation tests between the two were performed using the Spearman test. P<0.05 was considered statistically significant. The main objective of the study was to estimate the AUC of the ROC curve of ΔDicpleth and ΔPI to track IOH during induction. The sample size was determined with an expected AUC of 0.85, an expected incidence of hypotension of 80%, and a confidence interval width of 1. With a power of 80%, the number of patients to be included was then 62.16. The secondary objective was to estimate the AUC of the ROC curve (AUROC) for the combination ΔDicpleth and ΔPI. The statistical analysis was performed using Prism 6.00© (Graphpad Software, Inc, La Jolla, Calif., USA) and R 3.3.0 (R foundation for Statistical Computing, Vienna, Austria). Patients who had a non-measurable Dicpleth at baseline prior to induction of anesthesia were excluded from the analysis.

(97) Results

(98) From November 2014 to May 2015, 65 patients were included in the study. Prior to induction of anesthesia, Dicpleth was not measurable in 4 patients (6.2%) due to the absence of a detectable dicrotic notch on the PPG signal (class IV waveform according to Dawber et al.). Most patients had ASA II, with a mean age of 54 [39; 64] years. Hypertension, smoking and dyslipidemia were the most common comorbidities. The reason for neuroradiology procedures was mainly due to an aneurysm or an arteriovenous malformation with programmed embolization.

(99) Change in MAP, Dicpleth and PI During the Induction Period

(100) The median duration of induction of anesthesia was 11 [10; 13.5] minutes. A total of 720 “hemodynamic data points” were recorded: 61 at baseline and 659 after anesthetic injection, representing 659 changes from baseline.

(101) The MAP reference value was 86 [79; 93] mmHg, giving an individual IOH limit of 69 [62; 74] mmHg. MAP decreased to 54 [48; 60] mmHg before laryngoscopy and increased to 72 [64; 82] mmHg after tracheal intubation. The mean MAP over the overall induction period was 70 [64; 71] mmHg. Fifty-four patients (88%) had at least one IOH episode during induction of anesthesia, representing 323 measurements (49% of hemodynamic points). Twenty-eight patients (46%) received a bolus of vasopressors during induction (2 phenylephrine and 26 norepinephrine).

(102) Dicpleth was 0.54 [0.45; 0.65] at baseline, decreased to 0.36 [0.19; 0.45] (p<0.001) and increased to 0.46 [0.41; 0.56] after tracheal intubation. Baseline PI values were 1.7 [0.9; 3] and increased to 4.4 [2.8; 6.6] (p<0.001) before laryngoscopy and decreased to 3.6 [2.1; 5.4] after tracheal intubation. A visual representation of ΔMAP, ΔDicpleth and ΔPI during the induction period is described in FIG. 10. 89% agreement was found between ΔDicpleth and ΔMAP and 90% agreement between ΔPI and ΔMAP (ESM1).

(103) Diagnostic Performance of ΔDicpleth and ΔPI for the Detection of Hypotension

(104) The diagnostic performance values of ΔDicpleth and ΔPI are summarized in Table 1.

(105) TABLE-US-00001 TABLE 1 ΔDic.sub.pleth ΔPI ΔDic.sub.pleth + ΔPI AUC ROC 0.83 95% 0.86 (95% 0.91 (95% CI 0.80-0.86) CI 0.80-0.86) CI 0.88-0.95) P-value <0.001 <0.001 <0.001 Threshold value −19% 51% NA Sensitivity (%) 79 82 84 Specificity (%) 84 74 84 PPV (%) 79 71 79 NPV(%) 84 85 89

(106) ΔDicpleth: relative change in Dicpleth versus baseline; ΔPI: relative change in perfusion index versus baseline; AUC ROC: area under the receiving operative curve; PPV: positive predictive value; NPV: negative predictive value.

(107) The best cut-off values from ΔDicpleth and ΔPI for IOH detection were −19% and 51%, respectively. The AUCs of ΔDicpleth and ΔPI were not significantly different (p=0.22). The combination of ΔDicpleth and ΔPI to detect episodes of intrauterine homeostasis improved detection performance with an AUC the ROC curve (0.91, (95% Cl 0.88-0.95, p<0.001) statistically better than ΔDicpleth and ΔPI separately (p=0.026 and p<0.001, respectively).

(108) Change in MAP, Dicpleth and PI During Vasoconstrictive Administration

(109) Twenty-eight patients (46%) received a bolus of vasopressors during induction (2 phenylephrine and 26 norepinephrine). After vasopressors, DAP increased from 59 [50; 67] mmHg to 76 [68; 79] mmHg (relative change: 30% [14; 45], p<0.001). The number of people with diabetes increased from 0.34 [0.25; 0.39] to 0.48 [0.35; 0.55] (relative change: 44% [17; 63], p<0.001) and the PI decreased from 4.0 [3.3; 5.4] to 3.2 [1.8; 5.4] (relative change: −28% [−44; 13], p<0.001). ΔDicpleth and ΔPI under the effect of vasopressors were strongly related to ΔMAP (r=+0.73, 95% Cl 0.48-0.87, p<0.001 and r=−0.62 95% Cl −0.81 to −0.32, p<0.001; respectively).

(110) Change in Dicpleth and Dicradial During Vasoconstrictor Administration

(111) During maintenance of anesthesia, 48 boluses of norepinephrine were administered to 10 patients (5 [4; 6] boluses per patient) under invasive arterial pressure monitoring. Dicpleth was not measurable at 2 hemodynamic points which were excluded from the analysis. MAP increased from 70 [63; 77] mmHg to 88 [77; 98] mmHg (relative change 26% [19; 34], p<0.001). Dicpleth increased from 0.28 [0.17; 0.36] to 0.39 [0.25; 0.46] and Dicradial from 0.32 [0.21; 0.39] to 0.40 [0.31; 0.49] (relative changes 34% [20; 71], p<0.001 and 27% [14; 46], p<0.001, respectively). Dicpleth and Dicradial and their relative variations were highly correlated during vasoconstrictive administration (r=0.87 95% Cl 0.83-0.90 and r=0.92 95% Cl 0.85-0.95).

(112) These results show that Dicpleth can be used as a surrogate endpoint for non-invasive and continuous MAP monitoring during induction of anesthesia. These results show a strong correlation between ΔDicpleth and ΔMAP under the action of vasoconstrictors. A 19% drop in Dicpleth had good performance in detecting IOH with a sensitivity of 79% and a specificity of 84%. PI represents the ratio between the pulsatile and continuous components of light absorption. The results show a negative correlation between ΔPI and ΔMAP under vasopressors. ΔPI was also accurate for detecting IOH, but probably a little less accurate than ΔDicpleth for providing information on IOH intensity.