| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
Department of Anesthesiology, University of Florida College of Medicine, Gainesville, Florida
Address correspondence to Gregory M. Janelle, MD, Box 100254, Gainesville, FL 32610. Address e-mail to gjanelle{at}anest.ufl.edu. No reprints are available from the author.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In this study, we report on a beat-to-beat, noninvasive BP monitoring device, the T-Line® Tensymeter (Tensys Medical, Inc., San Diego, CA), used in a validation study that tested the hypothesis that the tensymeter-generated pressure waveform would provide real-time data comparable with that generated by an invasive arterial catheter system.
Arterial tonometry, which has been used for years as a noninvasive method of estimating arterial BP, involves the coupling of a patient's arterial pulsation with a pressure sensor. The pulsation can then be converted into a displayable waveform, and accompanying pressures can be computed using conventional bedside monitors. The challenge to obtaining accurate pressure measurement by tonometry is to find the optimal position of the pressure sensor over the artery and apply the sensor with enough force so that the pressure changes in the artery are accurately measured. This challenge is especially difficult, because the conditions that provide the best signal will change as the vessel tone and BP changes (2,3). A dynamic approach to optimizing the signal is therefore required if this method is to be clinically useful.
The T-Line® Tensymeter uses a new approach to tonometry that includes dynamic and continuous arterial location methods as well as novel signal-processing methods (dynamic applanation) for tracking BP variation. After the clinician applies the tensymeter sensor over the radial artery, the sensor is actively moved by an electromechanical system to an optimal lateral position using sensed pressure pulsations to optimize the position of the sensor over the radial artery. Location is maintained continuously using a servo-controlled scheme that forces the pressure sensor to return to the optimal location over a range of 17.1 mm through automatic lateral adjustments based upon error signal detection from optimal conditions. The tensymeter employs an applanation servo-control scheme that continuously monitors the peak-to-peak signal amplitude and determines the optimal applanation position. The monitoring method also uses a scheme (a variant of the pseudo-random binary sequencing technique) (4) for continuously varying the applanation position of the sensor using a statistically randomized method to insure that the pulse amplitude maximum is always determined. The tensymeter then incorporates the correction dependent on the patient's body mass index (BMI) via height and weight to provide for internal dynamic pressure waveform scaling and calibration. The goal of the study was to determine whether this new method provides estimates of BP that agree with invasive BP measurement within acceptable limits.2
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
10 mm Hg, and 4) subject weight not less 40 kg or more than 180 kg to satisfy BMI limitations in the device's calibration algorithms. All patients were studied after the induction of anesthesia, when the tensymeter was placed over the radial artery contralateral to the arm with the radial artery catheter. All radial artery catheters were 20-gauge in diameter and 4.8 cm long (B-D Insyte, Sandy, UT) and were connected to a transducer (Abbott Laboratories, Abbott Park, IL) by a 122-cm, low-volume, noncompliant length of tubing after careful de-airing. The quality of the waveform was evaluated for a sharp upslope and downslope of the arterial waveform. The natural frequency and damping coefficient of the pressure-monitoring system was evaluated using a standard flush test, ensuring a natural frequency of at least 20 Hz and damping coefficient of at least 0.2 (5,6). The tensymeter system consists of a disposable sensor that is placed over the wrist at the point where the radial artery is palpable over the head of the radius. The disposable component is secured to the patient with adhesive and then mated to a nondisposable wrist extension device to secure the electromechanically driven actuator to the sensor. This assembly is attached to a freestanding display monitor that can simultaneously display both arterial catheter and tensymeter tracings (Fig. 1). Otherwise, the tensymeter can be interfaced to display on the anesthesia monitor. The pulse amplitude for each beat is incorporated along with the patient's height and weight for pressure determinations and, although measurements of MBP are not affected by height and weight, the BMI is used to scale SBP and DBP before display. The system is somewhat motion-tolerant, in that a motion-detection algorithm is used to detect artifacts, resulting in a brief period (approximately 45 s) of off-line readjustment to relocate and applanate to the maximum pulsatile impulse. The version of the device tested was the TL-100.
|
The study was conducted in the operating room on a variety of adult surgical patients at an academic teaching hospital. Invasive arterial catheters and tensymeter sensors were both placed by care providers of variable experience. Each tensymeter application was supervised by one of three experienced users. Time to placement and data acquisition were recorded for both devices to evaluate the ability of a user to position the tensymeter on the patient relative to the time required to place an arterial catheter, expressed as mean ± sd and compared using Student's t-test. Data were simultaneously recorded from both the tensymeter and arterial catheter, and zero-referenced to the external auditory meatus for supine patients and to the cranial midline for patients in a lateral position. For the purposes of data collection, the arterial catheter signal from the pressure transducer was electronically "split" so the signal could be displayed on the local bedside monitor and simultaneously recorded on the analysis computer with the tensymeter pressure signal. The tensymeter processes all of its pressure information in real-time during active monitoring, independent of any external logging that may accompany its use during clinical research. This is true for both the tensymeter data and the arterial catheter pressure data. Beat detection operates simultaneously in real-time on both pressure signals wholly within the TL-100. SBP, DBP, and MBP are computed, displayed, and sent as data packets over an Ethernet link to the host computer for recording. The computational accuracy of the measurement algorithms was validated during development using simulator inputs over the entire range of physiologic BPs, with variable waveform characteristics, heart rates, and typical cardiac dysrhythmias. Likewise, clinical validation of the performance of the monitor's algorithms was performed in conjunction with the device's initial regulatory approval processes.
The host computer reads the log file and extracts "tagged" data packets representing the beat-by-beat numbers; the host reads these numbers as originally determined by the TL-100 during the actual run of the case without modification by any external agent. After logging, the individual beat-to-beat measurements for both the arterial catheter and tensymeter were passed through a seven-beat median filter designed to eliminate periods during which one or both monitors were not sampling actual patient BPs, such as during arterial catheter blood withdrawal, flushing, or during periods of off-line tensymeter recalibration.
Because arterial catheter blood sampling or flushing was not specifically recorded during the monitoring periods, it was detected by scanning the beat detect log file for a sudden (within 1 beat) shift in the arterial catheter pressure signal, either positively or negatively, that forced the signal level to a maximum or minimum for that monitoring period (e.g., SBP changes >100 mm Hg). This was interpreted to be an extraneous, nonphysiologic influence; therefore, beat comparisons with the tensymeter for any such beats were eliminated from computation.
Likewise, noise signals were detected when sudden, one- to two-beat rapid changes that exceeded 50 mm Hg in one measurement only (e.g., SBP) were detected. Any such detections were also visually verified by replaying the entire recording sequence (which, incidentally, is an additional validation feature of the log file recording capability of the system). Thereafter, paired differences were calculated for each simultaneous measurement. The paired error values were grouped into periods of 3 minutes, starting at the 5th minute of monitoring and every 10th minute thereafter during the session. Paired errors due to catheter flushing or other extraneous motion were eliminated from computation. Thus, over a 1-h recording period, 21 min of data were sampled. The paired differences, during these 3-min sample periods, were grouped into a sequential 10-beat window, or block averages. The monitoring period for comparison of tensymeter values with arterial catheter values was arbitrarily limited to 3 hours.
BP values generated from arterial catheters were used as controls. Using the 10-beat block averages, bias and precision were calculated for SBP, DBP, and MBP values. In addition, computations were made for the percentage of the 10-beat tensymeter averages that lay within 5, 10, 15, and 20 mm Hg of the corresponding arterial catheter values. Data were plotted individually on Bland-Altman plots.
To determine whether the device's dynamic and continuous arterial location methods and signal processing methods could compensate for changes in applanation over time or whether the continuous application of the tensymeter over the study limb would induce significant temporal inaccuracies because of compressive changes in the subcutaneous tissue or from unknown factors, the sampled recordings were subsequently analyzed for beat-to-beat comparison over the entire monitoring period. Specifically, calculations for the mean error of MBP, SBP, and DBP were made for each 10-min period of recording. Each 10-min period was then plotted and analyzed using time series analysis. Conventional descriptive statistics were also computed for these data.
The sample size of 25 patients was chosen based on recognized standards, by both the Food and Drug Administration and the American National Standards Institute, for evaluation of any noninvasive BP monitor.2 The Association for the Advancement of Medical Instrumentation (AAMI) SP10 specifies that a minimum of 15 subjects and 10 readings per subject should be reported if a contralateral arterial line is used as the reference. With 15 subjects, the 95% confidence interval would be 4 mm Hg, with sd of 8 mm Hg as the maximum allowable by the standard. With 25 subjects, the 95% confidence interval is 3.1 mm Hg. The substantially larger number of beat-to-beat measurements enhances the precision of the measurements beyond the AAMI recommendations (150 readings).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Fifteen patients were male (60%), and 10 were female (40%). The mean subject age was 61 ± 15 yr (range, 21 to 81 yr). The mean subject height was 170 ± 10 cm (range, 150 to 183 cm). The mean subject weight and BMI were 83.9 ± 18.6 kg and 27.9 ± 6.1 kg/m2, respectively (ranges, 52.3 to 120 kg and 20.0 to 42.7 kg/m2, respectively). The subjects' ASA physical status classifications were ASA 2 (n = 4), ASA 3 (n = 11), and ASA 4 (n = 10). Of eight subjects undergoing cardiac surgery, two had an intraaortic balloon pump in place. Other operations were thoracic aortic surgery (n = 1), abdominal aortic surgery (n = 1), orthopedic surgery (n = 5), craniotomy (n = 2), laparotomy (n = 7), and urological surgery (n = 1). Comorbid conditions included hypertension (60%), diabetes (24%), coronary artery disease (32%), congestive heart failure (20%), obesity (20%), and peripheral vascular disease (8%). Four of the 25 patients were in the lateral decubitus position; the remainder were supine.
The mean time to signal acquisition was significantly less for tensymeter versus arterial catheter (4.7 ± 2.5 vs 9.3 ± 9.1 min, respectively; P < 0.005). Simultaneous data were recorded for an average of 85.7 min per patient (range, 9.7–181.2 min), and an average of 148 ± 85 10-beat epochs per patient were analyzed.
A total of 17,009 measurements were available for paired analysis in the 25 patients studied. Fewer than 3% of all epochs were eliminated because of noise, arterial-line flushing, processing, or the seven-beat median filter. The mean ± sd bias and precision were as follows: 1.7 ± 7.0 and 5.7 ± 4.4 mm Hg for SBP, 2.3 ± 6.9 and 5.7 ± 4.5 mm Hg for DBP, and 1.7 ± 5.3 and 4.0 ± 4.8 mm Hg for MBP (Table 1). The range of pressure measurements was 41 to 189 mm Hg for SBP, 30 to 109 mm Hg for DBP, and 38 to 140 mm Hg for MBP. Bland-Altman plots and 95% confidence limits for 17,009 differences of tensymeter and arterial catheter measurements are shown in Figures 2–4 for SBP, DBP, and MBP and demonstrate that whereas the majority of tensymeter data points were within 5 mm Hg of arterial pressure measurements (67%), agreement was within 15 mm Hg in 94.6% or more of all measurements. Temporal analysis yielded a mean difference of mean BP of 0.2 mm Hg at baseline, peaking at 0.5 mm Hg over time (range, –0.4–0.5 mm Hg- Fig. 5). Mean differences of SBP and DBP peaked at 0.8–1.3 mm Hg, respectively, over time. There were no adverse events associated with use of the tensymeter system.
|
|
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Although it was not the purpose of this study to ascertain perfusion distal to the study device, it is important to note that there was no observed interference with any pulse oximetry devices applied to digits on the extremity ipsilateral to the tensymeter.
Likewise, although the tensymeter positioning device might be expected to interfere with performance of IV access cannulae, particularly those located at the dorsum of the ipsilateral wrist, it has been our experience that careful positioning circumvents this potential problem.
There are some limitations to the tensymeter. One patient in our study (enrolled before a change in the anesthetic plan) underwent a carotid endarterectomy under regional anesthesia with minimal sedation and was excluded because of limb motion. All other patients on whom the device was placed were included in the data analysis. Continuous or frequent patient limb motion currently limits the application of the studied version of the device (TL-100) to settings in which such motion artifact is absent (e.g., patients under general anesthesia, deep sedation, or cooperative awake patients). Although all of the patients in this study were predominantly in sinus rhythm, data were included from eight cardiac surgical patients (two with intraaortic balloon pumps) throughout the prebypass period, including periods of marked hemodynamic instability, such as during venous cannulation and initiation of cardiopulmonary bypass. These data, however, were not excluded from analysis.
Extensive intraoperative fluid resuscitation, although not explicitly studied, appears unlikely to significantly affect tensymeter performance. BMI is linearly related to weight and, although weight may change during surgery, the impact of such a change does not significantly affect system accuracy for SBP and DBP and does not alter measurement of MBP. Specifically, 17 weight bins and six height bins can be selected independently. The mean (meaning the value actually used in the computation) of each weight bin is the geometric mean of the lower and upper value, whereas the height bins are a simple average. Because the height does not change intraoperatively, BMI cannot be affected by the most influential factor (BMI varies as the inverse square of the height); thus, intraoperative weight gain is not clinically relevant to the tensymeter algorithm.
Although the majority of tensymeter data points were within 5 mm Hg of arterial BP measurements, and agreement was within 15 mm Hg in 94.6% or more of all measurements, the comparative data indicate that the tensymeter was not universally accurate throughout a portion of the monitored epochs. There are a number of potential explanations for this finding. Differences in arterial compressibility due to arterial wall stiffness from chronic hypertension, diabetes, or peripheral vascular disease could explain some differences. Likewise, initial calibration of the device is internal; therefore, minor malpositions of the device during this calibration phase can allow for suboptimal applanation of the sensor to the maximal arterial pulsation. Likewise, minor positional alterations due to movement of the patient or of the target artery underlying the subcutaneous tissue or imperfections in the dynamic applanation algorithm could induce temporal abnormalities. Temporal analysis, however, yielded acceptable clinical correlation.2 Changes in the arterial waveform on either the study limb or with the arterial catheter (from catheter obstruction, introduction of air into the monitoring tubing, peripheral vasoconstriction, or other unknown factors) may have been to blame. Although the number of data points used for comparison was large, the patient size and demographics did not allow for further elucidation of the underlying reasons for the BP discrepancies noted.
Use of the device is not feasible in patients with unidentifiable or congenitally absent radial arterial pulses. It is not recommended for use in patients with poor demonstrable collateral perfusion to the hand or in patients weighing less than 40 kg or more than 180 kg because of restrictions in the device's BMI categories used to calibrate and scale individual waveforms. The device requires only minimal expertise in locating the optimal application point at the maximal radial pulsation over the head of the radius, as the micromotor in the actuator has a 17.1-mm lateral pulse search-optimizing capability.
There are several other study limitations as well. The sample size of 25 patients for this initial study was relatively small; however, both the sample size and the number of data points included for comparison of the tensymeter and arterial catheter BP exceed the minimum recommended by the AAMI SP10 standard.2 Data were included from all monitoring periods, and these periods varied depending on the limitations induced by the surgical procedure itself. For this reason, subjects had variable monitoring periods, thus introducing possible bias in that there was the potential for subjects with better correlation to have longer periods of monitoring or, conversely, for patients with poorer correlation to have longer monitoring periods. It was felt that, rather than introducing investigator bias by arbitrarily choosing equal monitoring periods from each procedure for the purpose of comparison, all periods would be included in the data analysis.
In addition, signal processing and use of a 7-beat median filter were used to exclude data from comparison when both monitors were not functioning. This occurred during arterial sampling and during periods after excessive device movement, thus necessitating off-line recalibration. Although fewer than 3% of all epochs were eliminated because of noise or arterial line flushing, it is conceivable that these epochs contained periods of time during which the tensymeter yielded poor results compared with arterial catheter data. The exclusion of arterial catheter data from sudden alterations in BP (e.g., within one heartbeat, SBP changes >100 mm Hg) makes it unlikely, although not impossible, that such exclusions were erroneous. Likewise, data were omitted because of noise, such as that which results from sudden movement of the tensymeter, and were limited to periods in which the SBP changed by 50 mm Hg or more over 1 to 2 heartbeats; these changes were not only automatically detected by the monitoring computer but also visually verified by replaying the epoch in question. As must be acknowledged, no attempt to correlate patient or limb motion with the periods excluded because of noise was attempted. This method introduces the possibility that the tensymeter monitoring system failed to obtain reliable BPs (for <3% of monitoring periods). We believe that this finding instead illustrates the susceptibility of the tensymeter monitoring system to such noise as a potential weakness of the device. Conscious of this limitation, the manufacturers have subsequently marketed a second-generation device, the TL-150, that reportedly offers such improvements as less motion susceptibility and more rapid motion recovery.
There are three currently marketed BP monitoring devices that offer noninvasive alternatives to arterial cannulation for frequent BP determinations. The Finapres (TNO Biomedical Instrumentation, Amsterdam, The Netherlands) measures capillary bed pressure of fingers based on the Peñáz volume-clamp method (9,10). It has been shown to inconsistently correlate with arterial BP, (11–13) and the technique has been associated with distal tissue hypoxia (14). Although this device is no longer marketed, Finapres Medical Systems currently offers the Finometer and Portapres devices based on similar technology (Finapres Medical Systems BV, Arnhem, The Netherlands). The Colin N-CAT tonometer (Colin Corporation, Komaki, Japan) requires a cuff to calibrate and is therefore rendered susceptible to inaccuracies from calibration itself, including errors associated with rhythm disturbances such as atrial fibrillation (15). This device has been reported to be less accurate than the Finapres, particularly at lower MBP (12). The Medwave Vasotrac (Medwave, Arden Hills, MN) does not offer truly simultaneous pressure data. This device averages 12 to 15 beats and reports updated pressure intervals; thus, it is neither beat-to-beat nor real-time (16). Despite this methodological difference, Belani et al. (16) published a study similar to this one in which the Vasotrac also exceeded AAMI standards with respect to bias and precision compared with direct radial artery measurements in more than 17,000 measurement in 80 patients.
In summary, continuous BP monitoring is frequently desirable in subjects with cardiac or cerebrovascular disease, in critically ill patients, in surgical procedures in which periods of rapid blood loss are likely, and any case in which hemodynamic instability is anticipated.3 In this first validation study, continuous BP measurements obtained with the T-Line® Tensymeter agreed with simultaneous NIBP measurements by an arterial catheter within an acceptable range for the population of surgical patients studied. The results indicate that the tensymeter provides a noninvasive alternative for continuous arterial BP measurement and functions within AAMI specifications for BP measurement.2 The tensymeter may enable physicians to circumvent arterial cannulation in some patients when beat-to-beat BP measurement is desirable. Future studies are warranted to determine the applicability of the TL-100 (or its successor, the TL-150) to a number of different clinical scenarios, such as those involving massive fluid resuscitation, in patients with unstable cardiac rhythms (e.g., atrial fibrillation), awake (or sedated) patients, and in patients with significant peripheral vascular disease.
|
Footnotes |
|---|
2 American National Standard ANSI/AAMI SP10:2002. Manual, electronic, or automated sphygmomanometers. Arlington, Virginia: Association for the Advancement of Medical Instrumentation; 2003: pages 30–57.
3 American Society of Anesthesiologists: ASA Standards, Guidelines and Statements; Statement of Invasive Monitoring Procedures, October 15, 2000.
Financial support provided by Tensys Medical, Inc., San Diego, California and the Department of Anesthesiology, University of Florida, Gainesville, Florida.
Presented in part at the Annual Meeting of the American Society of Anesthesiologists, San Francisco, CA, October 11–15, 2003 and in part at the International Anesthesia Research Society 78th Clinical and Scientific Congress, March 27–31, 2004.
Conflict of Interest: Dr. Janelle participates in bi-annual compensated meetings of the Tensys Medical, Inc. medical advisory board.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
