Many limitations of manual intermittent blood pressure measurement have been overcome by automated ..... Careful analysis of the arterial pressure waveform.
0
Faculties of Medicine and Applied Sciences INSTITUTE BIOMEDICAL TECHNOLOGY IBITECH HYDRAULICS LABORATORY GHENT UNIVERSITY, BELGIUM
EVALUATION OF ARTERIAL TONOMETRY IN A CLINICAL SETTING by Alain Kalmar
Promoter: Prof. dr. ir. P. Verdonck Co-Promoter: ir. K. Matthys Thesis submitted to obtain the degree in the specialized studies of biomedical and clinical engineering
Academic Year 2002-2003
1
Acknowledgement This thesis could never have been realised without the help of many whom where involved more or less directly during the passed year. I’d like to thank a few persons in particular, which I am very grateful: Prof. dr. ir. P. Verdonck, for his conscientious follow-up of the progress during the entire academic year; Ir. K. Matthys, for technical support and advice and for helping me with the software processing of the data; Prof. dr. E. Mortier, Prof. dr. J. Van Aken and Prof. dr. M. Struys for the moral and practical support in the planning and execution of the data acquisition, and help in administration; Ir. F. van Thuyne and Mr. D. De Guytere for the technical support in the adjustment of the electronic connections and mechanical accessories; Sophie Moriaux, for the moral support in more difficult times; I also wish to thank my parents, without whose help this wouldn’t have been possible. Finally I’d wish to thank everybody who was always interested in the progress of my work, even dough I didn’t see them often because of this busy year.
Alain Kalmar Gent, June 2003
2
Contents Acknowledgement ...................................................................................................................... 1 Contents ...................................................................................................................................... 2 1
Introduction ........................................................................................................................ 4
2
Literature overview ............................................................................................................ 5 2.1
Heart rate monitoring ................................................................................................. 5
2.2
Pulse rate monitoring ................................................................................................. 5
2.3
Arterial blood pressure monitoring ............................................................................ 5
2.3.1
Indirect measurement of arterial blood pressure ................................................ 6
2.3.1.1
Manual intermittent techniques ...................................................................... 6
2.3.1.2
Automated intermittent techniques ................................................................ 7
2.3.1.3
Automated continuous techniques.................................................................. 9
2.3.1.4
Applanation tonometry ................................................................................. 11
2.3.2
Direct measurement of arterial blood pressure................................................. 12
2.3.2.1
Percutaneous radial artery cannulation. ....................................................... 13
2.3.2.2
Considerations about pressure waveforms ................................................... 13
2.3.2.3
Complications of direct arterial pressure monitoring ................................... 14
2.3.2.4
Technical aspects of direct blood pressure measurement ............................ 14
2.3.2.5
Frequency content of the arterial pressure waveform .................................. 14
2.3.2.6
Natural frequency and damping coefficient ................................................. 15
2.3.2.7
Adequate dynamic response ......................................................................... 16
2.3.2.8
Clinical measurement of natural frequency and damping coefficient .......... 17
2.3.2.9
Pressure monitoring systems ........................................................................ 18
2.3.2.10 Transducer setup: zeroing, calibrating and levelling ................................... 18 2.3.2.11 Arterial pressure waveforms ........................................................................ 19 3
Study objectives ............................................................................................................... 20
4
Methods ............................................................................................................................ 21 4.1
Bracelet design ......................................................................................................... 22
4.2
Electronic connection ............................................................................................... 23
4.3
Data acquisition ........................................................................................................ 24
4.4
Data computation ..................................................................................................... 25
4.5
Data interpretation and selection in Excel ................................................................ 26
4.6
Visual interpretation of tonometric and invasive pressure signals in Excel............. 28
3 4.6.1
Evaluation of short-term accuracy of tonometric signal .................................. 28
4.6.2
Evaluation of long-term accuracy of tonometric signal ................................... 32
4.7
Data export to Matlab. .............................................................................................. 37
4.8
Evaluation with calibration in Excel, short-time ...................................................... 40
4.8.1
Reliability during hemodynamic stability. ....................................................... 41
4.8.2
Reliability during a fast pressure change. ........................................................ 44
4.8.3
Reliability during a slow pressure change. ....................................................... 45
4.8.4
Conclusions on short-time reliability of the calibrated tonometric signal. ...... 48
4.9
Assessment of the accuracy of the calibrated signal over a total acquisition........... 49
5
Tonometric artefacts caused by cuff inflation .................................................................. 56
6
Discussion & Conclusion ................................................................................................. 61
Appendix 1 : Advice from the Ethical Committee ................................................................... 64 Appendix 2 : Technical specifications ..................................................................................... 72 Appendix 3 : Visual Basic programming in Excel ................................................................... 73 Appendix 4 : Visual evaluation of tonometric signal ............................................................... 79 Appendix 5 : Total acquisition comparison ............................................................................. 83 Appendix 6 : Influence of mean pressure on cuff-artefact ....................................................... 88 References ................................................................................................................................ 89
4
1 Introduction Arterial tonometry is a technique that allows continuous and non-invasive registration of the arterial pressure waveform. Being non-invasive, it also means a fast and simple approach and above all, with only a small load for the patient compared to an interventional pressure measurement. Applanation tonometry is based on a pressure transducer placed upon a superficial artery, which is supported by bone structure so that adequate applanation of the vessel can be achieved to obtain a pressure waveform. As applanation tonometry can only provide for the waveform, this signal has to be calibrated by means of an external method, for example a cuff technique giving minimum and maximum blood pressure values, allowing the waveform to be rescaled to its correct offset and amplitude. The aim of this work is to acquire tonometric and invasive pressure waveforms and to evaluate feasibility of deducing an approximation of invasive pressure based on tonometric pressure waveform. In order to acquire necessary data, specific hardware and software had to be developed or modified. During standard operative procedure, continuous invasive pressure monitoring, continuous tonometric pressure monitoring and intermittent oscillometric cuff-pressure monitoring were stored. In a subsequent off-line analysis of the data, different algorithms were applied on tonometric data in order to approximate invasive pressure. Different approaches are discussed to improve tonometric registration, signal quality, signal transformation, calibration and user comfort.
5
2 Literature overview Many standard physical means of assessing circulation are employed throughout an operation. For many healthy patients undergoing minor procedures, these physical signs may provide a considerable fraction of the total cardiovascular monitoring. The most obvious method for assessing circulation is palpation of the pulse. Also, during cardiac surgery the beating heart may be observed directly and palpation of the ascending aorta by the surgeon provides a useful estimate of central blood pressure. When the operation becomes more complex or the patient comes to surgery with more advanced or unstable cardiovascular disease, the extent of supplemental electronic monitoring grows accordingly.
2.1 Heart rate monitoring In modern practice, electronic monitoring devices are used to provide a continuous, numeric display of heart rate. The most common technique applied in the operating room is electrocardiography. The digital value displayed for heart rate is generated from an algorithm designed to count and average a certain number of beats and then display a number that is updated every 5 to 15 seconds1. Despite multiple measures and averaging filters, the monitor may display inaccurate heart rates when the ECG trace is distorted by patient movement or other electrical interference. In the operating room, this frequently occurs when the electrosurgical unit is in use. In these instances the arterial pressure trace or the pulse oximeter plethysmograph can be an alternative though indirect source of information about heart rate.
2.2 Pulse rate monitoring It might be argued that monitoring pulse rate is more important than monitoring heart rate, in terms of peroperative hemodynamic assessment. The pulse oximeter plethysmograph trace will provide a suitable pulse measurement source for most patients except those with severe arterial occlusive disease or those with marked peripheral vasoconstriction. Automatic noninvasive blood pressure devices determine the pulse rate by counting oscillations in pressure sensed by the surrounding cuff. When direct arterial pressure measurement is in place, the pressure waveform provides a reliable pulse source. However, one must be aware of possible misleading by non-systolic arterial pulsations as can occur in patients treated with intra-aortic balloon counter pulsation or aortic valve regurgitation. In contrast, patients with pulsus alternans may have inappropriately low pulse rates measured, owing to the diminished magnitude of every other arterial pulsation.
2.3 Arterial blood pressure monitoring As with heart rate, blood pressure is a fundamental cardiovascular vital sign, which reflects the force that drives perfusion of the body. In addition, blood pressure is the most important determinant of left ventricular afterload, the workload of the heart. Techniques for measuring blood pressure fall into two major categories: indirect Riva-Rocci cuff devices and direct arterial cannulation and pressure transduction. These methods differ in nearly every respect, notably in terms of the physical process being monitored and in the level of invasiveness of their application. In clinical practice, blood pressures
6 measured by different techniques often yield significantly different values2. Studies comparing different blood pressure monitoring techniques usually use direct arterial pressure measurement as the reference standard against which another method is judged. However, there are many ways in which direct arterial pressure measurement can yield spurious results3. 2.3.1 Indirect measurement of arterial blood pressure
2.3.1.1 Manual intermittent techniques Most indirect methods of blood pressure measurement rely on a Riva-Rocci sphygmomanometer. As described by Riva-Rocci in 1896, this apparatus included an arm-encircling inflatable elastic cuff, a rubber bulb to inflate the cuff, and a mercury manometer to measure cuff pressure. Riva-Rocci described the measurement of systolic arterial blood pressure by determining the pressure at which the palpated radial arterial pulse disappeared as the cuff was inflated4. A variation of the Riva-Rocci method commonly employed today is generally termed the return-to-flow technique. Whereas the Riva-Rocci technique recorded the pressure during cuff inflation at which the pulse completely disappeared, the return-to-flow method records the pressure during cuff deflation at which the pulse reappears. Although return-to-flow methods provide a simple, rapid means to estimate systolic blood pressure, they do not allow measurement of diastolic blood pressure. Undoubtedly, the most widely used intermittent manual method for blood pressure determination is the auscultation of sounds originally described by Korotkoff in 19055. Using a Riva-Rocci sphygmomanometer and cuff, Korotkoff applied a stethoscope to the artery directly below the cuff to auscultate the sounds generated as the cuff was slowly deflated. These sounds are a complex series of audible frequencies produced by turbulent flow, instability of the arterial wall, and shock wave formation created as external occluding pressure on a major artery is reduced3. The pressure at which the first Korotkoff sound is auscultated is generally accepted as the systolic pressure (phase I). The sound character progressively changes (phases II and III), becomes muffled (phase IV), and finally absent (phase V). Diastolic pressure is recorded at phase IV or V. However, phase V may never occur in certain pathophysiologic states, such as aortic regurgitation6. The auscultatory method for blood pressure determination is limited by excessively long or loose stethoscope tubing, which impairs sound transmission, or by poor hearing sensitivity of the observer. Aneroid (=spring-gauge) manometers are subject to calibration errors and should be checked periodically. A more basic shortcoming of auscultation is the reliance on blood flow to generate Korotkoff sounds. Pathologic or iatrogenic causes for decreased peripheral blood flow, such as cardiogenic shock or high-dose vasopressor infusion, can attenuate or obliterate sound generation and result in significant underestimation of blood pressure7. In contrast, low compliance of the tissues underlying the cuff, as encountered in a shivering patient, will require an excessively high cuffoccluding pressure and produce "pseudo-hypertension."3 Patients with severe calcific arteriosclerosis may have relatively noncompressible arteries. Another circumstance wherein cuff blood pressures will yield spuriously elevated results compared with true intra-arterial blood pressure.
7
Other common sources of error during intermittent manual blood pressure measurement include selection of an inappropriate cuff size or excessively rapid cuff deflation. Blood pressure cuff width should be 20 percent greater than arm diameter, and the cuff should be applied snugly after any residual air has been squeezed out. The pneumatic bladder inside the cuff should span at least half the circumference of the arm and be centered over the artery. Cuffs that are too narrow yield Figure 1 : Effect of cuff size on manual blood pressure erroneously elevated values for blood measurement. An inappropriately small blood pressure cuff erroneously high values for blood pressure because pressure8,9. Cuff deflation rate is another yields the pressure within the cuff is incompletely transmitted to important variable that influences accurate the underlying artery. blood pressure measurement, especially when deflation is performed manually. The decrease in cuff pressure should proceed slowly enough for the Korotkoff sounds to be auscultated and assigned properly to the current pressure in the cuff. Failure to identify the initial Korotkoff sounds will result in a falsely low measure of blood pressure. A deflation rate of 3 mm Hg/s limits this source of error, and coupling deflation rate to heart rate—2 mm Hg/beat—has been found to improve accuracy further10.
2.3.1.2 Automated intermittent techniques Many limitations of manual intermittent blood pressure measurement have been overcome by automated non-invasive blood pressure devices (NIBP), which are now used widely in medical care. By applying a single algorithm or method of data interpretation, NIBP devices provide consistent, reliable values for systolic, diastolic, and mean arterial pressure (MAP). Most automated NIBP devices are based on the technique termed oscillometry, a technique first described by von Recklinghousen in 1931.11 In this method, variations in cuff pressure resulting from arterial pulsations during cuff deflation are sensed by the monitor and used to determine arterial blood pressure values. Peak amplitude of arterial pulsations corresponds closely to true MAP12,13. Values for systolic and diastolic pressure are derived using proprietary formulas that examine the rate of change of the pressure pulsations (Fig 2) 14. Systolic pressure is generally chosen as the pressure at which pulsations are increasing and are at 25 to 50 percent of maximum. Diastolic pressure is more difficult to determine but is commonly placed at the point where the pulse amplitude had declined by 80 percent3.
8
Figure 2 : Comparison of blood pressure measurements using Korotkoff sounds and oscillometry. Oscillometric systolic blood pressure is recorded at the point where cuff pressure oscillations begin to increase, mean pressure corresponds to the point of maximal oscillations, and diastolic pressure is measured where the oscillations become attenuated.
Because the oscillometric NIBP technique provides accurate pressure measurements in paediatric patients15,16, several investigators have proposed using neonatal sized cuffs placed around a finger or thumb of an adult patient17. Although these alternatives may be appropriate in certain circumstances, their overall accuracy has not been widely validated and may not conform to accepted industry standards18. Many other techniques have been described for automated NIBP measurement, but none have supplanted the Standard oscillometric technique. One device uses the Doppler principle to determine blood flow distal to the cuff19 and another senses motion of the arterial wall20. Although these have been found to record blood pressure with acceptable accuracy, they require extra efforts to place and stabilize additional sensing transducers. Supraorbital artery oscillometry has been studied as an alternative site for blood pressure measurement but appears inaccurate compared with direct radial artery pressure values, perhaps owing to the peripheral location and vasoreactive sensitivity of the supraorbital artery21. Photo-oscillometry to detect brachial artery movement has also been described, but awaits broader validation in clinical practice22. Under controlled clinical conditions, numerous investigators have demonstrated that automated NIBP measurements closely approximate directly measured arterial pressure2,8,13. However, other studies underscore the fact that marked disagreement occurs when direct and indirect pressure measurements are compared, particularly when radial artery pressure is used as the direct measurement standard23,24 or when techniques are compared under changing clinical conditions. When direct brachial artery pressures have been compared with various indirect methods, (including manual auscultation, automated oscillometry, aneroid manometer, visual
9 onset of needle oscillations and return to flow), the relation between the indirect and direct pressures varied between patients, within patients over time, and with changing hemodynamic conditions25. As noted, some authors have emphasized the lack of exact agreement between different measurements of blood pressure26,22. Standards for performance of automated NIBP devices have been advanced by organizations such as the American Association for the Advancement of Medical Instrumentation (AAMI) and the British Hypertension Society. AAMI standards require a monitor to record blood pressure to within 5±8 mm Hg (mean ± Standard deviation) prediction error compared with the reference method22. However, clinical performance of an NIBP monitor should be evaluated by other criteria as well. These include the number of outlier values, duration of discrepancies, magnitude of individual errors, and performance under variable clinical conditions22.
2.3.1.3 Automated continuous techniques In the past, continuous blood pressure monitoring required direct arterial cannulation. Advances in microprocessor and servomechanical control technology have enabled noninvasive techniques to provide a reasonable representation of the arterial pressure waveform and a nearly continuous assessment of blood pressure. The most widely investigated of these devices measures finger blood pressure using a servoplethysmomanometer, designed and first reported by Penaz in 197327. Many systems have been manufactured since. One commercial model, the PORTAPRES®28 is a battery-operated portable instrument to monitor finger arterial pressure continuously. In essence, the finger blood pressure device uses an arterial volume-damp method. It consists of a small cuff secured around the middle phalanx of a finger or the base of the thumb. The inflatable, flexible cuff contains an infrared photoplethysmograph. The cuff and photoplethysmograph are linked through a servo-controlled mechanism housed within a small box that is strapped to Figure 3 : Portapres system the wrist. The plethysmograph continually measures the size (i.e., diameter) of the digital arteries using transillumination. To begin monitoring, a "locking" calibration procedure is performed by varying cuff pressure to establish the vessel size at which oscillometric pressure variation is maximal. As noted, this corresponds to MAP. An electromechanical feedback loop is then established, and external pressure applied to the cuff is varied continuously to keep the measured vessel size constant at the set-point. Thus, cuff pressure tracks arterial pressure throughout the cardiac cycle and is displayed on the monitor screen as a continuous waveform. Another commercial model, Colin 7000 (Colin medical instruments Corp, San Antonio, Texas, US), is the combination of a cuff-oscillometric device, approximately ten centimetres proximally to the wrist, and a tonometric device, approximately three centimetres
10 proximally to the wrist. Blood pressure is measured intermittently by oscillometric method (semiconductor sensor, 1-300 mmHg), while continuous blood pressure waveforms are provided by tonometric means (semiconductor sensor array, 250 Hz, 0-300 mmHg).
It consists of a sensor-array placed on the wrist over the radial artery. This sensor contains piezoelectric pressure transducers separated by 0.2mm. A pneumatic pump and bellows press the transducer array against the skin and tissue above the artery. This pressure is known as the hold down pressure (HDP). To determine optimal hold-down pressure, the monitor searches through a range of pressure values until it measures a signal indicating that the artery is of the form shown (Fig 4). When the artery is partially flattened, a graph, called a tomogram, can be plotted Figure 4 : determining optimal hold-down pressure to show sensor pulse amplitude versus transducer number. The individual sensor elements whose pulse amplitudes are near the maximum pulse amplitude are calibrated to the systolic and diastolic values obtained in the oscillometric cuff measurement. The system has digital (BP, PR; serial, RS-232) and analog (BP waveform; 0-5V) output. A Third commercial model is the Finapress®. This portable recording system that utilizes a technique called "vascular unloading" to provide continuous measurements of pulse rate and blood pressure with a cuff around the middle finger. It can be attached to a computer for ongoing storage of blood pressure waveforms.29 This device seems to be out of use, since no recent publication of it was found in literature. (Ohmeda, Englewood, Colorado, USA)
Continuous non-invasive finger blood pressure measurement devices have undergone numerous clinical evaluations comparing their performance with direct arterial pressure measurements30. Many of these investigations have demonstrated small overall mean differences between finger and intra-arterial pressure measurements. However, small overall mean pressure differences do not necessarily indicate good measurement agreement. In clinical practice, the frequency of major measurement errors may be the more relevant and important issue. For example, Smith et al found that finger arterial spasm precluded accurate pressure measurement in 5 percent of patients30. By definition, finger blood pressure monitoring records a distal arterial pressure, which tends to be lower than brachial arterial pressure in elderly patients with atherosclerosis and to be higher than brachial pressure in young patients, because of peripheral pulse wave amplification. Using sophisticated electronic processing techniques, Bos et al31 demonstrated that the pulse wave shape and pressure values of the finger blood pressure signal could, by using a frequency-dependent transfer function, be corrected to resemble direct brachial artery pressure waveforms within the limits of accuracy described by the AAMI. However, these sophisticated algorithms have not yet been integrated into commercially available monitors.
11 The Portapres® finger cuff blood pressure measurement system seems to provide accurate data in a wide range of clinically relevant blood pressures, even in patients with shock, dysrhythmia, or severe hypertension. Because the "transducer" recording blood pressure with this method is the finger, the vertical height of the finger becomes an important determinant for the pressure recorded. Finally, the potential for circulatory impairment of the distal finger caused by the constantly inflated cuff has been a cause for concern. Gravenstein et al32 demonstrated mild hypoxemia in the capillary blood of the fingertip during finger blood pressure monitoring. No adverse outcomes were noted in these study patients or in others in whom finger blood pressure measurement was performed for as long as 7 hours. Despite the apparent safety of these devices for short-term use, these considerations have limited more widespread clinical application of continuous finger blood pressure monitoring. Other automatic and continuous techniques have been used to measure blood pressure noninvasively. One such device reconstructs an arterial pressure waveform from arterial wall displacement measurements, following an oscillometric calibration. Evidently, changes in arterial compliance affect the clinical performance of this instrument33. Because arterial pulse-wave velocity depends on arterial blood pressure, another device uses pulse transit time as recorded from dual pulse oximeter probes placed on the ear and finger following oscillometric calibration from the contralateral arm34. This device has also performed poorly in clinical settings, with indirectly monitored blood pressures changing in an opposite direction from the direct intra-arterial pressure more than 30 percent of the time.
2.3.1.4 Applanation tonometry A third method used for continuous non-invasive pressure monitoring is arterial tonometry, a version of applanation tonometry35,36. Arterial applanation tonometry is a technique that allows continuous and non-invasive registration of the arterial pressure waveform. Such a continuous registration offers a pressure-time wave morphology that cannot be assessed by conventional cuff methods (Korotkoff/auscultation or oscillometry). The principle of applanation tonometry is illustrated in Figure 5: a pressure transducer is placed upon a superficial artery, which is supported by bone structure so that adequate applanation of the vessel can be achieved to obtain the pressure waveform. As applanation tonometry can only provide for the waveform, this signal has to be calibrated
Figure 5 : Calibration via an occlusive cuff method.
12 by means of an external method, for example a cuff technique giving minimum and maximum blood pressure values, allowing the waveform to be rescaled to its correct offset and amplitude. For a better understanding, we have to address the socalled Imbert-Fick principle. The Imbert-Fick law states that the internal pressure (Po) in a spherical body with an infinitely thin, dry and elastic membrane wall, equals the force (W) exerted on this body divided by the applanation surface (A) (Figure 6)37. Hence, in an applanated Spherical body, the Imbert-Fick law can be written as: Pt = W/A, and if one can measure W an A, Pt can be derived. Figure 6 : Imbert Fick principle
In brief, a superficial artery (usually the radial) is compressed and partially flattened against the underlying bone. This flattened arterial surface serves as a "transducer" for intravascular pressures acting perpendicularly against the vessel wall. An array of piezoelectric crystals positioned on the skin overlying this flattened portion of artery senses arterial pressure changes and translates them into a continuous arterial pressure waveform. The device is calibrated at intervals by cuff oscillometry from the upper arm. Early investigations suggested that the clinical performance of this device was better than other forms of continuous non-invasive pressure monitoring. Unfortunately, more recent clinical studies have identified limitations of this device in paediatric patients38 and in patients receiving vasodilating drugs22. Like other automated, continuous non-invasive blood pressure techniques, arterial tonometry holds promise as a means of rapidly detecting changes in blood pressure. However, in terms of absolute accuracy of blood pressure measurement, intermittent oscillometry appears to be superior to radial artery tonometry. 2.3.2 Direct measurement of arterial blood pressure Arterial cannulation with continuous pressure transduction and waveform display remains the accepted standard for blood pressure monitoring. Indications for and advantages of direct arterial pressure measurement can be considered to fall into four major categories: 1. Continuous real-time monitoring. Direct arterial catheterisation is chosen when rapid moment-to-moment blood pressure changes are anticipated, and their immediate detection is considered to be a high priority. The Australian Incident Monitoring Study of 1993 confirmed the superiority of direct arterial pressure monitoring over indirect monitoring techniques for the early detection of intraoperative hypotension39. 2. Failure of indirect blood pressure measurement. Any cuff-based measurement technique may prove impossible in some morbidly obese patients or those with burned extremities. 3. Supplementary diagnostic clues. Careful analysis of the arterial pressure waveform can provide many additional subtle pathophysiologic insights40. Some important examples are described below. 4. Arterial blood sampling. When frequent samples of arterial blood are required, an arterial catheter provides reliable vascular access and obviates the need for multiple arterial punctures.
13
2.3.2.1 Percutaneous radial artery cannulation. Several peripheral arteries are available for percutaneous cannulation, but radial artery pressure monitoring is most common in anaesthesia and critical care. It is technically easy to perform and rarely associated with complications, owing to good collateral circulation of the hand41. Slogoff et al described 1700 cardiovascular surgical patients who underwent radial artery cannulation without ischemic complications, despite evidence of radial artery occlusion after decannulation in more than 25 percent of patients42. Although a 20-gauge Teflon catheter is the most common one used for radial artery cannulation, the size (20 or 18-gauge) and composition (Teflon or polypropylene) of the catheter seem to have little influence on the frequency of complications. Several techniques are used for placing an arterial catheter; the arterial cannulation using an integrated needleguidewire-catheter assembly is commonly used: - The wrist is positioned and the artery identified by palpation. - The needle is introduced through the skin and advanced toward the artery, generally at 30 to 45 degrees. - A flash of arterial blood into the collection reservoir identifies the artery as the needle tip enters the vessel. - The guide wire is advanced through the needle into the vessel lumen. - The catheter is advanced over the guide wire. - After the catheter is fully advanced into the vessel lumen, the guide-wire is removed and a narrow bore, low compliance pressure tubing is fastened to the catheter. - The apparatus is securely fixed to the wrist and connected to the pressure transducer.
2.3.2.2 Considerations about pressure waveforms If the radial arteries are unsuitable for pressure monitoring, several alternative cannulation sates are available: the ulnar artery, brachial artery, dorsalis pedis, posterior tibial, superficial temporal arteries, axillary artery and femoral artery. As with axillary arterial pressure monitoring, the femoral artery waveform more closely resembles aortic pressure than waveforms recorded from peripheral sites. Several factors influence the choice of site for arterial pressure monitoring, even though the radial artery of the nondominant hand is usually preferred for most applications. One important consideration is whether significant regional differences in blood pressure exist. Frank et al43demonstrated that 21 percent of patients undergoing peripheral vascular surgery had a blood pressure difference between the two arms that exceeded 20 mmHg. During hypothermia, thermoregulatory vasoconstriction causes radial arterial systolic pressure to exceed femoral artery systolic pressure44. In these patients, radial arterial pressure underestimates central aortic pressure, often by more than 20 mmHg45. Anyway, one must recognize that even under normal circumstances, the arterial pressure waveform changes its morphology as it is transmitted through the vascular tree, resulting in a
14 wider pulse pressure recorded from a peripheral artery than from a more central location 46. These normal physiologic differences in arterial pressure recorded from different sites should be considered whenever direct arterial pressure monitoring is performed.
2.3.2.3 Complications of direct arterial pressure monitoring Large clinical investigations confirm a low incidence of long-term complications following radial artery cannulation, in particular, the small risk of distal ischemia, which is likely in the range of 0.1 percent or less47. It is striking that many if not most of the complications from direct arterial pressure monitoring can be attributed to equipment misuse48. Serious complications, although rare, do occur following arterial cannulation. Reports describe retained guide wires requiring surgical extraction49, fatal haemorrhage following difficult femoral artery cannulation50, and upper extremity compartment syndrome following brachial artery cannulation51. Infectious complications are increasingly rare now that disposable transducers have replaced reusable ones.
2.3.2.4 Technical aspects of direct blood pressure measurement The displayed pressure signal is influenced by the system plumbing, including the arterial catheter, extension tubing, stopcocks, flush devices, transducer, amplifier and recorder. Direct pressure monitoring systems used in medical practice are described as underdamped secondorder dynamic systems52. These systems exhibit a behaviour depending on three characteristic physical properties: elasticity, mass and friction. These three properties determine the system operating characteristics, termed the frequency response or dynamic response, which in turn is characterized by two important system parameters, natural frequency (fn, ) and damping coefficient (, Z, , D). Natural frequency describes how rapidly the system oscillates, and the damping coefficient describes how rapidly it comes to rest.
2.3.2.5 Frequency content of the arterial pressure waveform The arterial blood pressure waveform is a periodic complex wave, which can be reproduced by Fourier analysis, a technique that recreates the original complex pressure wave by summing a series of simpler sine waves of various amplitudes and frequencies. The original pressure wave has a characteristic periodicity termed the fundamental frequency, which is equal to the pulse rate. The sine waves that sum to produce the complex wave have frequencies that are multiples or harmonics of the fundamental frequency. A crude arterial waveform, which displays a systolic upstroke, systolic peak, dichotic notch and so forth, can be reconstructed with reasonable accuracy from
Figure 7 : Arterial blood pressure waveform produced by summation of sine waves. The fundamental wave (top) added to 63% of the second harmonic wave (middle) results in a pressure wave (bottom) that resembles an arterial blood pressure waveform.
15 two sine waves, the fundamental frequency and the second harmonic (fig 7)40. If the original arterial pressure waveform contains high-frequency components such as a steep systolic upstroke or other sharp details, higher frequency sine waves (and more harmonics) are needed to provide a faithful reconstruction of the original pressure waveform. As a general rule, six to ten harmonics are required to provide adequate reproduction of most arterial pressure waveforms53. Hence, accurate blood pressure measurement in a patient with a pulse rate of 120 beats/min (2 cycles/s or 2 Hz) requires a monitoring system dynamic response of 12 to 20 Hz. Clearly, the faster the heart rate and the steeper the systolic pressure upstroke, the greater will be the dynamic response that is required from the monitoring system.
2.3.2.6 Natural frequency and damping coefficient If the monitoring system has a natural frequency that is too low, frequencies in the monitored pressure waveform will approach the natural frequency of the measurement system. As a result, the system will resonate, and pressure waveforms recorded on the monitor will be exaggerated or amplified versions of true intra-arterial pressures. (Fig.8)40 This phenomenon is the familiar arterial pressure waveform that displays overshoot, ringing, or resonance. Tachycardia and steep systolic pressure upstrokes present the greatest challenge for clinical monitoring, because the higher frequency contents of these waveforms more likely approach the resonant frequency of the measurement system.
Figuur 8 : Underdamped arterial pressure waveform. Systolic pressure overshoot and additional small, nonphysiologic pressure waves (ar-rows) distort the waveform and make it hard to discern the dicrotic notch (boxes). Digital values displayed for direct arterial blood pressure (ART 166/56, mean 82 mm Hg) and noninvasive blood pressure (NIBP 126/63, mean 84 mm Hg) show the differences in pressure measurement that arise because of an underdamped arterial pressure waveform.
The bedside monitoring system must not only have a sufficiently high natural frequency, but it must also have an adequate damping coefficient. The overdamped arterial pressure waveform has a slurred upstroke, absent dicrotic notch, and loss of fine detail. Severely overdamped pressure waves display a falsely narrowed pulse pressure, although the MAP value may remain reasonably accurate. In contrast, underdamped pressure waveforms display systolic pressure overshoot and contain additional artefacts that are produced by the measurement system but not part of the original intravascular pressure waveform (Fig. 8). They are produced by an underdamped monitoring system that continues to ring or oscillate abnormally in response to the input pressure signal.
16
2.3.2.7 Adequate dynamic response Most catheter-tubing transducer systems are underdamped but have an acceptable natural frequency that exceeds 12 Hz. If the system natural frequency is lower than 7.5 Hz, the pressure waveform may be distorted, and damping adjustment cannot make the monitored waveform resemble the original waveform adequately54. Conversely, if natural frequency can be increased sufficiently (e.g. 24 Hz), damping will have minimal effect on the monitored waveform, and faithful reproduction of intravascular pressure is achieved more easily. In other words, the lower the natural frequency of the monitoring system, the more Figuur 9 : Interaction between damping coefficiënt and natural Depending on these two system parameters, catheter narrow will be the range of damping frequency. tubing-transducer systems fall into one of five different dynamic coefficients that can be tolerated to ensure response ranges. Systems with an optimal dynamic response will record the most demanding pressure waveforms, whereas those with faithful pressure wave reproduction or faithfully adequate dynamic response will record accurately most pressure adequate dynamic response. For example, if waveforms seen in clinical practice. Overdamped and underdamped introduce artifacts characteristic of these technical limitations. the monitoring system natural frequency is systems Systems with a natural frequency less than 7 Hz are considered 10 Hz, the damping coefficient must be unacceptable. The rectangular crosshatched box indicates the ranges of coefficients and natural frequencies commonly encountered between 0.45 and 0.6 for accurate pressure damping in clinical pressure measurement systems. The point within the box waveform reproduction. In this situation, too shows the mean values of 30 such systems recorded by Schwid. low a damping coefficient causes the system to be underdamped, to resonate, and to produce a factitiously elevated systolic blood pressure. In contrast, too high a damping coefficient produces an overdamped system, in which systolic pressure is falsely decreased and fine detail in the pressure trace is lost (fig 9)40,55. In summary, a pressure monitoring system will have optimal dynamic response if its natural frequency is as high as possible. In theory, this is achieved best by limiting the length of tubing and using only stiff tubing that is designed for pressure monitoring56. Blood clots and air bubbles contained within the tubing-stopcock transducer plumbing will adversely affect the dynamic frequency response in a similar fashion. So it is best to keep the monitoring system simple with the fewest components necessary. Devices can be added to increase damping without lowering natural frequency 40. These devices work through impedance matching, which eliminates wave reflections and prevents resonance in the monitoring system57. Limitations of these devices include the inability to adjust or tune the monitoring system to provide the most accurate in vivo pressure recordings58.
17
2.3.2.8 Clinical measurement of natural frequency and damping coefficient The method used most often to evaluate the dynamic response of the monitoring system is the fast flush test59. Under most conditions, this method yields results that are essentially identical to those from the standard laboratory square-wave testing60; it tests the entire monitoring system, from catheter tip to transducer. To perform this test, the fast flush valve is opened briefly several times, and the resulting flush artefact is examined. Monitoring system natural frequency is related to the time period or distance between two successive oscillation cycles. An example is illustrated (fig 10)40, where the distance between two cycles is 1.7 mm at a recorder speed of 25 mm/s. Natural frequency is then easily calculated: 1 cycle/1.7mm x 25 mm/s = 14.7 cycles/s or 14.7 Hz. Note that the distance between two successive oscillation cycles should be identical. Furthermore, the tighter the oscillation cycles, the higher the natural frequency will be. To measure natural frequency most accurately, a fast recording speed should be chosen (e.g. 50 mm/s), several flush cycles should be examined, and an average value calculated. The damping coefficient is determined from the flush artefact by measuring the amplitudes of Figure 10 : Fast Flush method successive oscillation cycles. The amplitude ratio so derived indicates how quickly the measuring system comes to rest. A low amplitude ratio corresponds to a high damping coefficient, or a system that comes to rest quickly. Conversely, a high amplitude ratio corresponds to a low damping coefficient, or a system that tends to resonate. The damping coefficient can be calculated mathematically, but it is usually determined graphically from the measured amplitude ratio61. In this example (fig 10)40, the amplitudes of two successive oscillation cycles are 24 and 17 mm, respectively, giving an amplitude ratio of 17/24 or 0.71, which corresponds to a damping coefficient of 0.11 using the graphic solution shown. As in the case for measuring natural frequency, any two adjacent peaks may be used to determine the amplitude ratio, because the ratio of successive peaks should be relatively constant. Again, this is a fundamental characteristic of the measurement system - the damping coefficient. In an attempt to optimise the dynamic response of the monitoring system response of the monitoring system, one should attempt to keep the catheter-tubing-transducer system as simple as possible, using the minimum length of tubing and number of stopcocks required for patientcare purpose. Secondly, the fast-flush method allows calculating natural frequency and damping coefficient.
18 Owing to the dynamic response limitation of most clinical pressure-monitoring systems, direct measurement of systolic arterial pressure often exceeds indirect non-invasive measurement, simply because of underdamping and resonance (Fig 9) 58.
2.3.2.9 Pressure monitoring systems Arterial pressure monitoring systems have a number of components beginning with the intra-arterial catheter and including extension tubing, stopcocks, continuous flush device, pressure transducer, and bedside monitor with a waveform display screen. The flush device provides a continuous, slow (1-3 ml/h) infusion of saline to purge the monitoring system, which thereby prevents thrombus formation within the arterial catheter. The monitoring system also includes a spring-loaded valve that allows periodic, high- Figuur 11 : invasive pressure monitoring device pressure flushing to purge the extension line. The stopcocks in the system provide sites for blood sampling and allow the transducer to be exposed to atmospheric pressure to establish a zero reference value.
2.3.2.10
Transducer setup: zeroing, calibrating and levelling
Prior to initiating patient monitoring, the pressure transducer must be zeroed, calibrated, and levelled to the appropriate position on the patient. The initial step in this process is to expose the transducer to atmospheric pressure by opening the adjacent stopcock to air, pressing the zero pressure button on the monitor, and thus establishing the zero pressure reference value. The transducer now has a reference - ambient atmospheric pressure against which all intravascular pressures are measured. This process underscores the fact that all pressures displayed on the monitor are referenced to atmospheric pressure, outside the body. At this point, it is the air-fluid interface at the level of the stopcock that is the zero pressure locus. This point must be aligned with a specific position on the patient to ensure the correct transducer level. If - after some time - the stopcock is reexposed to atmospheric pressure and pressure is not equal to zero, baseline drift of the transducer’s electrical circuit may have occurred. These technical problems appear to be uncommon now that high-quality disposable transducers are widely available62. Currently, disposable pressure transducers meet or exceed accuracy standards established by the AAMI and the American National Standards Institute62. Transducer calibration at the bedside thus appears to be unnecessary. In general, if a pressure transducer or monitoring cable is faulty, the initial zero value cannot be established, and the monitoring system must be changed. Rarely, despite successful zeroing, the recorded pressure values appear erroneous, and a malfunctioning pressure transducer, cable, or monitor must be suspected and replaced63.
19 The final step in transducer setup is levelling the pressure monitoring zero point to the appropriate position on the patient. In the supine patient, pressure transducers are levelled most often to the midchest position in the midaxillary line, a site chosen because it is easy to estimate by eye and provides a reasonable approximation for the midpoint of the heart in the chest64.
2.3.2.11
Arterial pressure waveforms
Direct arterial pressure monitoring in anaesthetized patients began more than 50 years ago65. The systemic arterial pressure waveform results from ejection of blood from the left ventricle into the aorta during systole, followed by peripheral arterial runoff of this stroke volume during diastole (fig 12)40. The systolic components follow the ECG R wave and consist of a steep pressure upstroke, peak and decline and correspond to the period of left ventricular systolic ejection. The downslope of the arterial pressure waveform is interrupted by the dicrotic notch, then continues its decline during diastole following the ECG T wave, reaching its nadir at end-diastole. The dicrotic notch recorded directly from the central aorta is termed the incisura. The incisura is sharply defined and related to aortic valve closure. In contrast, the peripheral arterial waveform generally displays a later, smoother dicrotic notch that only approximates timing of aortic valve closure and depends more on arterial wall properties.
Figuur 12 : Normal arterial blood pressure waveform and its relation to the electrocardiographic R wave (1) systolic upstroke, (2) systolic peak pressure, (3) systolic decline, (4) dicrotic notch (5) diastolic runoff, en (6) enddiastolic pressure.
As the arterial pressure wave travels from the central aorta to the periphery, several characteristic changes occur. The arterial upstroke becomes steeper, the systolic peak becomes higher, the dicrotic notch appears later, the diastolic wave becomes more prominent, and the end-diastolic pressure becomes lower. Thus, compared with central aortic pressure, peripheral arterial waveforms have higher systolic pressure, lower diastolic pressure, and wider pulse pressure. Furthermore, there is a delay in the arrival of the pressure pulse at peripheral sites, so that the systolic pressure upstroke begins approximately 60 milliseconds later in the radial artery than in the aorta. Despite morphologic and temporal differences between peripheral and central arterial waveforms, the MAP in the aorta is just slightly greater than that in the radial artery66.
20 Pressure wave reflection is the predominant factor that influences the shape of the arterial pressure waveform as it travels peripherally. As blood flows from aorta to radial artery, mean pressure only decreases slightly, because there is little resistance to flow, but then falls markedly in the arterioles, owing to the dramatic increase in vascular resistance at this site. This high resistance to flow diminishes pressure pulsations in small downstream vessels but acts to augment upstream arterial pressure pulses owing to pressure wave reflection67. Multiple studies68 underscore the importance of wave reflection in determining the shape Figuur 13 : Impact of pressure wave reflection on arterial of the arterial pulse recorded from all sites pressure waveforms. In elderly individuals with reduced arterial in the body, in health and disease. For distensibility, early return of reflected waves increases pulse pressure, produces a late systolic pressure peak (arrow), and attenuates the example, elderly patients have reduced diastolic pressure wave. arterial dispensability, which results in early return of reflected pressure waves, an increased pulse pressure, a late systolic pressure peak, and disappearance of the diastolic pressure wave (Fig. 13)69. Multiple pathologic conditions have specific features on the morphology of arterial pressure waveforms. Of special interest is the influence of mechanical ventilation on arterial pressure waveforms. Large cyclic changes in systolic blood pressure occur in many patients who are mechanically ventilated. Systolic pressure variation is the difference between maximal and minimal values of systolic arterial pressure recorded over the mechanical positive-pressure respiratory cycle70,71.
Figure 14 : Systolic pressure variation. Compared with systolic blood pressure recorded at end expiration (1) a small increase occurs during positive pressure inspiration (2) followed by a decrease (3). Normally, total systolic pressure variation does not exceed 10 mmHg. In this instance, the large down indicates hypovolemia even though systolic arterial pressure and heart rate are relatively normal.
3
Study objectives
The purpose of this study is to determine if non-invasive continuous blood-pressure measurement based on tonometric assessment of a radial artery pressure waveform is accurate enough to be - in some cases - an alternative for invasive radial artery blood pressure measurement. Alternatively, can it be a useful extension of classical intermittent cuffmeasurement?
21
4
Methods
Four consecutive adults, scheduled for major neurosurgical operation were enrolled in this prospective study. Written informed consent and institutional approval were obtained (see Appendix 1). Exclusion criteria were subclavian stenosis and pre-existing radial artery cannulation. Routine continuous monitoring in the operating room included electrocardiography, pulse oxymetry, capnography, non-invasive blood pressure, invasive blood pressure and rectal temperature. General anaesthesia was induced by intravenous propofol 2 mg/kg. Sis-atracurium 0.1 mg/kg was used for muscle relaxation. General anaesthesia was maintained by continuous infusion of propofol 6 mg/kg/h, remifentanil 0,1 g/kg/u and sis-atracurium 0,15 mg/kg/u. Patients were intubated orally and ventilated mechanically to achieve and-tidal CO2 and oxygen saturation within normal limits. After anaesthesia induction, a 20-gauge 8 cm PE catheter72 was inserted percutaneously into the left radial artery, 1 cm proximal to the wrist. The catheter was connected via 150 cm long (1.5 mm internal diameter) rigid pressure tubing, filled with saline to a continuous flush pressure-transducer system73. The system was calibrated against atmospheric pressure. The midaxillary line was used as the zero-reference point. At the right wrist, a piezo-electric applanation tonometer (Millar® Mikro-Tip® catheter transducer model SSD-936) was positioned on the right radial artery, 1 cm proximal to the wrist. Optimal location was determined by digital palpation and evaluating waveform on the anaesthesia monitor.
After locating the optimal sensor position, it was immobilized by means of tegaderm® patch. Optimal ‘hold down’ pressure was adjusted by means of a bracelet with screw. Also at the right arm, non-invasive oscillometric cuff blood pressure measurement was applied. It was configured at a sampling rate of 1/3minutes. All patients remained in the horizontal position. Patients were kept normothermic by a forcedair warming system.
22
4.1 Bracelet design The bracelet has to conform to a few properties: It has to be unstretchable. During fastening of the bracelet, the position of the sensor must not change. So it’s absolutely necessary to design a bracelet that can be fastened after its position with respect to the sensor is determined. During fastening of the bracelet, the sensor should not be pressed too hard to the artery, otherwise it’s impossible to obtain a partial flattening of the artery afterwards. After the bracelet is fastened enough, the backpressure on the sensor should be adjustable, in order to obtain a partial flattening of the artery so that a maximal amplitude of the pulse pressure can be obtained. It is dexterous that one can see and inspect the sensor during and after the backpressure is adjusted. Once the bracelet is fastened and backpressure is adjusted, the system should be rigid enough not to change spontaneously during at least an hour. The position of the sensor or bracelet, as well as the ‘hold down’ pressure should change minimally. It should be water-resistant, electrically isolating and not toxic or allergising, in order to maximise patient-safety. Designing a prototype of a bracelet, I tried to choose a mould and materials in order to comply with the desired properties as much as possible. I The system is fixed with two Velcro® tapes. This material is a polymer (Nylon 6.6), which is known not to induce allergic reactions, to be nonconductive and non-stretchable. After the bracelet is located over the sensor, it can be easily fixed by pressing the two tapes onto each other, without changing the position of the fixing piece anymore. Due to its non-stretching property, once fixed, the circumference of the bracelet doesn’t change anymore. The fixing piece is made of polycarbonate, which also is non-allergenic, non-conductive, nontoxic, and non-stretchable. An important advantage of the polycarbonate, its transparency, makes it possible to inspect sensor position and skin of the patient after the bracelet is positioned. Two clefts were milled for attaching the Velcro tapes. The velcro tapes were attached to the fixing piece by sewing them with a polyamide wire. This method also minimizes stretchability of the system and avoids use of reactive chemicals. In the piece, a cavity was made that is one millimetre broader than the sensor. Also a fissure of one millimetre broad and deep was made for placing the wire of the sensor. On top of the ‘sensor hole’, a screw hole is drilled in which a polyethylene screw is positioned. Polyethylene is also non-allergenic, non-conductive, non-toxic, and waterresistant. When screwed in the screw hole, it tenses just enough to be adjustable, but to stay immobile when not actively manipulated. The contact-properties of polyethylene and polycarbonate are suitable for this application. The top of the screw is blunted by a few millimetres, so the angle by which the sensor is pressed to the skin can be adapted a little bit, in order to optimise pulse pressure.
23
4.2 Electronic connection All monitoring equipment, including both sensors was connected to a Datex AS/3 monitor (Datex-Ohmeda Inc. Madison, WI). Because the Datex AS/3 monitoring system has no standard connection for tonometric sensors, a connection was made with a transducer-input cable. In this cable, four wires are present. The connection for the tonometer also has four wires. When connected to the monitor, two wires receive a DC excitation voltage of 5V. Optimal Excitation voltage for the tonometer is between 2.5 and 7.5 V DC, so no further adaptation had to be made and a simple connection could readily be made. On the two other wires, voltage is measured by the monitor. The monitor has an optimal range between –7 and +7 mV. Since sensitivity of the tonometer is 5 V/V/mmHg, this also falls in optimal range. (e.g. a blood pressure of 130 mmHg and an excitation voltage of 5V delivers a voltage of 3,25 mV) For technical description, see appendix 2.
Figure 15 : disposable pressure transducer cable
Figure 16 : tonometer connected to disposable cable
24
4.3 Data acquisition Collecting all data via only one integrated monitoring and computing system, has a lot of advantages: - Perfect synchronisation of multiple signals, - Maximal patient-safety since only approved equipment is in close contact with the patient and no extra power source is introduced. - Minimal use of complicated equipment, which is a major advantage in clinical situations. - Having all the data stored in one single synchronised file greatly facilitates interpretation afterwards. While the pressure-transducer system directly provides an accurate description of arterial pressure, tonometric data only give relative measurement. These data have to be calibrated by oscillometric data to provide beat-to-beat absolute pressure-values. Since the tonometer was connected to an input for another commercial pressure-transducer, the curves, shown on the monitor are only indicative for signal quality, but no absolute values can be derived from it. For translating the tonometric measurements to absolute pressure, further off-line transformation of the data have to be done afterwards. All data from the monitor were sampled via the Datex-Ohmeda Collect Software® package for subsequent off-line analysis. Invasive and tonometric pressure waveforms were sampled at 300 Hz. Non-invasive blood pressure measurement was recorded every 3 minutes.
25
Connection of the Datex-monitor to a laptop computer is done with a 9-Pin serial cable, which is plugged in the RS-232 output of the monitor and to the PC. Then, the collect software automatically recognizes available signals and can be adjusted to show the desired waves. A maximum of four waves can be visualised and stored simultaneously, while all available numerical data are recorded every five seconds. Once activated, the Collect® software stores the desired data into a single file. In subsequent off-line analysis, the data were transformed to an ASCII-file and imported in Excel for visual interpretation. Further, a program was written in Matlab™-environment (The Mathworks, Inc, Natick, MA, USA) in order to analyse calibration techniques of the acquired data.
4.4 Data computation Since data are recorded at 300 Hz and a few hours of recording were done in a single file, the data files are too large to be readily imported in Excel. Some Visual Basic (See Appendix 3) programming was necessary to cope with and to visualize the data. Since even fast modern computers have difficulties with manipulating these large data files, and ASCII-output from the collect software are not readily exploitable by Excel, the data had to be transformed first. An acquisition of three hours, with a sampling rate of 300 Hz and a collection of three waveforms (ECG, Invasive pressure, Tonometry) provides a file of three columns and 3,240,000 rows. Excel only can handle 64,000 rows, but it can handle up to 255 columns. For convenient manipulation of the data, I decided to store the data in one matrix of multiple columns. In order to have expedient visual access to subsequent data, I decided to store data into subsequent columns from column 4 on, while leaving column 1 to 3 free for copying columns of interest onto. This way, a graphical representation of subsequent curves was easily done by copying the desired columns onto column 1 to 3. Since in excel, a curve is attributed to a certain column, simply changing the content of the column changes the represented curve. Although 30,000 integers have to be copied and plotted, allows computing time to be less than one second, which greatly eases visually inspecting large data files.
26
4.5 Data interpretation and selection in Excel After processing of the data for making them accessible for visual interpretation, adequate samples could be chosen for further analysis.
Figure 17: tonometric pressure and invasive pressure during steady state. Visual representation in Excel.
In this example, the waves with greatest amplitude are invasive pressure waveforms; absolute values of the pressure can be derived from it, shown in ordinate. The waves with smallest amplitude are tonometric data, no absolute values can be derived from it, since calibration has to be made, based on cuff-values. Since pressure, registered by the tonometer is not only intra-arterial pressure, but the sum of intra-arterial pressure, vascular stiffness, tissue, and other factors, one must consider that a significant error might be introduced when other than intra-vascular sources of fast pressure change are not reckoned with. A remarkable phenomenon is shown, which could be explained by rising of tissue-pressure. When looking at the visual representation of the datasets, it is striking that, since cuffmeasurement is on the same arm as tonometry, every time the cuff inflates, the tonometry shows this - at first sight - paradoxal artefact. Since the shape of the artefact is repetitively very similar, there must be a physical explanation for this phenomenon:
Figure 18: tonometric and invasive registration during cuff-inflation. Cuff and tonometric sensor are on the same arm; invasive pressure catheter is on the other arm
One would expect the tonometric blood pressure measurement to diminish while a cuff is inflated proximally, but the opposite seems to occur. For further discussion of this phenomenon, see chapter 5.
27 This feature shows that other causes than arterial pressure can very strongly influence tonometric measurements. One can see that total pressure on the tonometric device more than doubles because of pressure sources other than arterial pressure in a very short period of time. Of course, one has to take into account these kinds of artefacts when interpreting tonometric registrations. Also, one could imagine that other factors could influence venous or tissuepressure that should be reckoned with when interpreting tonometric pressure-acquisition.
28
4.6 Visual interpretation of tonometric and invasive pressure signals in Excel 4.6.1 Evaluation of short-term accuracy of tonometric signal Visual interpretation of tonometric and invasive pressure signals, based on the curves in excel allows assessing whether tonometric and invasive signals divert in the same direction or not. For each acquisition of 10000 points (i.e. 33 seconds), four possible groups were distinguished: 1. Invasive and tonometric signal deviate in the same direction.
2. Invasive and tonometric signal deviate in an opposite direction, or one signal diverts while the other stays steady state.
3. Both signals show are stable, are in steady state.
4. Bad signal quality, no assessment of trend possible
Inability of interpretation because of cuff inflation was not taken into account.
29
Patient 1 Patient 2 Patient 3
Same direction Opposite direction 0 5 3 4 27 10
Steady-state 108 38 85
Bad signal 0 7 9
Number of windows (1 window = 10000 samples) corresponding to the group 1-2-3-4. For every patient most windows describe a steady-state situation, where tonometric and invasive signal have a very good correspondence. When pressiure variations occur, the deviation of the tonometric signal does not always follow the invasive signal.
Invasive pressure variation caused by patient ventilation is often adequately followed by tonometry. Only if the amplitude of these variations was high enough (>25%, as in following illustration), the signal was considered as deviating in the same direction, since in most cases these variations are of no hemodynamic significance.
Figure 19: influence of patient ventilation on blood pressure is nicely detected by tonometry. (Patient 2 at 1900 seconds).
Figure 20: fast changes in blood pressure are detected very well by the tonometer ( Patient 3, first acquisition at 3600 seconds)
As one can notice in the trends (see chapter 4.6.2.), fast and small variations of blood pressure are detected by the tonometer, but relatively slow (over tens of seconds) variations are not detected most of the time. Most probably, this incapacity to detect slow pressure variations adequately is not due to hardware failure, but rather a result of physiological changes in the arm as a result of the pressure change. In order to correct for these inaccuracies, additional sensors or algorithmic approaches could be considered.
30 In next curve, one can see a rather fast rise in invasive blood pressure signal and initially, a tonometric signal that rises in absolute values and amplitude. After 12 seconds, invasive curve continues to rise, but tonometric signal diminishes, until after 45 seconds, invasive signal is at a systolic pressure of 200 mmHg, while tonometric signal is at its initial values.
Figure 21: initially, tonometric signal nicely follows invasive signal, but at the end of the curve (total curve = 50 seconds), while systolic pressure is 200 mmHg, tonometric signal has retaken its initial values. (Patient 2 at 1800 seconds).
Another example of the tonometer’s inability to detect slow pressure changes is shown in following curve.
Figure 22: Two subsequent acquisitions, separated by a 550 seconds of slow blood pressure drop while tonometric signal remains roughly at steady-state. Patient 1 at 200 and 800 seconds.
In this figure, the first curve shows a stable blood pressure curve. The second curve acquired after a slow pressure drop over 550 seconds. Once more, although the small variations in blood pressure due to ventilation of the patient are nicely detected by the tonometer, a manifest lower blood pressure after a period of 550 seconds is not reproduced by the tonometric signal. Minute variations of blood pressure due to patient ventilation are of no clinical relevance (when considering only absolute blood pressure), but larger pressure drops are of much higher clinical interest. Considering this clinical fact, one must decide that in this patient-case the tonometer was not able to detect the relevant pressure change, even though the breathing pattern of the invasive signal is nicely followed by the tonometric signal. An analogous situation can be shown in patient 2 during a slow rise in blood pressure. The first curve is made at 7400 seconds over a period of 33 seconds; the subsequent curve is made after a slow pressure rise over a period of 133 seconds; the third curve is made after a further
31 slow rise in blood pressure over a period of 200 seconds. One can see that from the first to the second curve, tonometric signal tends to follow the invasive signal, but after the subsequent rise, the tonometric signal goes the opposite direction.
Figure 23: Three subsequent acquisitions separated by respectively 133 and 200 seconds during a slow rise in blood pressure. Patient 2 at 7400, 7566 and 7800 seconds.
Another illustration can be made from patient 3, first acquisition, where the first curve is made at 475 seconds over a period of 50 seconds; the second subsequent curve is made after a slow pressure drop over a period of 475 seconds, sharing opposite effect again, but in the other direction.
Figure 24 Two subsequent acquisitions, separated by a 475 seconds of slow blood pressure drop while tonometric signal remains roughly at steady-state. Patient 3, first acquisition at 475 and 1000 seconds.
Conclusion concerning short-term accuracy: In the above paragraph different hemodynamic events are illustrated along with different tonometric behaviours, among which : Effects of patient ventilation and other short-term pressure variations. The tonometric signal very accurately follows these short-term pressure ventilations. Different events where pressure changes over a longer time occur. In these cases, the deviation of the tonometric signal is not readily predictable.
32
4.6.2 Evaluation of long-term accuracy of tonometric signal When evaluating former curves, the impression rises that most of the time, the tonometric signal adequately detects fast pressure changes, but tends to return to an initial ‘optimal’ signal amplitude, more or les independent of invasive blood pressure over a period of time. In order to evaluate this hypothesis, trend curves are shown. For the total time of the acquisition, systolic, diastolic and mean pressures were determined for each heartbeat. These values are drawn at 0.5 Hz. Mean pressure was calculated for each point (i.e. every 2 seconds) with the formula: Mean = (systolic pressure + 2 * diastolic pressure) / 3.
Figure 25: Example of trend curve for systolic, diastolic and mean pressure of both invasive and tonometric pressure. This curve shows a total of 12 minutes of acquisition of patient 2.
In order to have a more interpretable representation, only mean values are shown in the following curves.
33
The following curves are a representation of mean arterial pressure (blue) and tonometric signal (purple) of the first patient. Each curve shows a total of 600 points, which equals 1200 seconds or 20 minutes.
The obvious repetitive rise of tonometric pressure at exactly every five minutes is due to cuffinflation at the brachial artery of the right arm. While the artefact is very characteristic and reproducible, subtle differences between distinct artefacts could possibly contain appealing information about hemodynamic conditions. One can clearly see that there is a poor relationship between changes in invasive and tonometric pressure-change. For further discussion of these cuff-artefacts, see chapter 5
34 The following curves are a representation of mean arterial pressure (blue) and tonometric signal (purple) of the second patient. Each curve shows a total of 1500 points, which equals 3000 seconds or 50 minutes.
In the first curve, the cuff-artefact occurs much more frequently. This is because of the more frequent inflation of the cuff by the algorithm when exceptional pressures are registered or problems during measurement occur.
35 The following curves are a representation of mean arterial pressure (blue) and tonometric signal (purple) of the third patient, first acquisition. Each curve shows a total of 1200 points, which equals 2400 seconds or 40 minutes.
36 The following curves are a representation of mean arterial pressure (blue) and tonometric signal (purple) of the third patient, second acquisition. Each curve shows a total of 1200 points, which equals 10 minutes.
37
4.7 Data export to Matlab. Once data are easily visualised, interesting datasets can be identified and transferred to the Matlab program for analysis of signal quality and calibration techniques of blood pressure signals. Of most interest are datasets where a significant rise or fall of blood pressure occurs, since the ability of the tonometer to detect and correctly quantify these changes is an absolute condition for its usefulness. A suitable dataset is chosen on basis of visual assessment in excel. In this case, we choose an acquisition of 30 seconds where a modest pressure drop occurs.
Figure 26: Visual assessment of Invasive and tonometric pressure waves in Excel in a period of 30 seconds
38 This dataset is exported to an ASCII file, modified to be readable in the Matlab program (see fig. 20).
Figure 27: Screenshot of the custom-written user interface in Matlab-environment to calibrate pressure signals and compare invasive with non-invasive recordings.
At this point, cuff values of diastolic and systolic pressure have to be delivered to the program so it can try to extrapolate invasive pressure out of tonometric values. Different ways of calibration are possible. In this study, three different strategies are compared; in every case the first cycle is calibrated with diastolic cuff-pressure and either systolic of mean cuff-pressure. 1. For each cycle, diastolic and systolic pressure are rescaled to cuff-value. Every drift in pressure is compensated. Evidently, this is not useful when real pressure changes occur and are to be detected. Although, it can be useful in very short datasets, when one knows pressure is not to change significantly. In these cases, we consider that every change in trend is an artefact. It can be useful to evaluate technical qualities of sensors and electronics, software or algorithms.
39
In this example, the recalibrated signal evidently shows a constant diastolic and systolic value. As the invasive pressure drops, there is an increasing discrepancy between invasive waveforms and calibrated tonometric values. 2. Another approach is to start calibrating only the first cycle by means of cuff values. In this case, diastolic and mean tonometric values are calibrated. For calculating the calibrated tonometer signals, diastolic tonometric value is translated to diastolic cuff value, and then gain is adjusted in this way that mean tonometric pressure corresponds to mean cuff value. The latter is either input data from the cuff device, or in case only systolic and diastolic cuff values are given, the mean cuff value is calculated with the standard formula: MAP=DBP+(SBP-DBP)/3. Using mean pressure instead of systolic pressure can be preferable since in distal arteries, mean values deviate less than systolic values. Since the cuff measurement is done at the brachial artery, and the tonometer is located on the radial artery, using mean values might be preferable. On the other hand, since invasive pressure is measured radialy, one could state that discrepancy between distal and proximal arteries is of no importance, since one compares with distal measurement. The obtained calibration values for the first cycle are then applied on all subsequent cycles. This method allows real pressure changes to be reproduced by tonometric measurement. Of course, other than intra-arterial sources of tonometric pressure change will be represented as supposed arterial pressure change.
In this case, since tonometric pressure rises while invasive pressure lowers, a rising discrepancy develops between measured and calculated arterial pressure. 3. In a third approach, the transformation is analogue to the second transformation applied above, but now, a new transformation is applied on every cycle in order to recalibrate diastolic and mean pressure values to cuff values, instead of applying the transformation of the first cycle on every cycle.
40
4.8 Evaluation with calibration in Excel, short-time These calibration methods each have their advantages and disadvantages, depending on the desired application. The most promising of these algorithms for evaluating changes in blood pressure during operative procedures seems to be the second one. This algorithm is then used in Microsoft Excel for further evaluation of a calibrated tonometric signal with an invasive signal in large datasets. Tonometric diastolic and systolic pressure of the first tonometric and invasive pulse of a sequence was assessed by an algorithm (see appendix 2). Then invasive diastolic and systolic values were used to calibrate the tonometric waveform. Using invasive pressure for calibration instead of cuff-values allows for more precise and quantitative evaluation of the reliability of the tonometric signal after pressure changes. Gain was defined as follows: Gain = Arterial Pulse Pressure / Tonometric Pulse pressure Gain =
Arterial Systolic Pressure – Arterial Diastolic pressure Tonometric Systolic Pressure – Tonometric Diastolic Pressure
For each point, Calibrated Tonometric Pressure (CTP) was defined as follows: CTP = Arterial Diastolic Pressure + [(Tonometric Pressure – Tonom.Diast. Pressure) * Gain] On one waveform the calibration method has following results:
Figure 28: Calibrated tonometric signal for one pulse from patient 1.
When applying this algorithm on subsequent pulses (but keeping calibration values constant), the following curves were obtained. Depending on degree of hemodynamic stability, different situations can occur:
41
4.8.1 Reliability during hemodynamic stability. Systolic and diastolic values of invasive pressure waves have a high degree of stability.
Figure 29: Patient 1, example 1. Calibrated tonometric signal in steady state. (total curve = 33 seconds)
The calibrated tonometric curve nicely follows the invasive pattern.
Figure 30: Patient 1 example 2. Calibrated tonometric signal in steady state. (total curve = 33 seconds)
The calibrated tonometric signal follows invasive pressure signal, though the influence of ventilation on the invasive pressure signal is followed by the tonometric signal with a delay of roughly 1 pulse. The arrows indicate that the pulse with maximal calibrated tonometric systolic pressure comes one pulse after the one with maximal invasive systolic pressure. The cause of this behaviour is unclear.
42
Figure 31: Patient 2 example 1. Calibrated tonometric signal in steady state at high blood pressure. The calibrated tonometric signal follows invasive pattern, even ventilatory variations are adequately reproduced. (total curve = 33 seconds)
Figure 32: Patient 2 example 2. Calibrated tonometric signal in steady state at normal blood pressure. The calibrated tonometric signal follows invasive pattern. Again ventilatory variations are adequately reproduced. (Total curve = 33 seconds)
Figure 33: Patient 3, Acquisition 1, example 1. Calibrated tonometric signal in steady state. The calibrated tonometric signal follows invasive pattern. (Total curve = 33 seconds)
43
Figure 34: Patient 3, Acquisition 1, example 2. Calibrated tonometric signal in steady state. The calibrated tonometric signal follows invasive pattern.
Figure 35: Patient 3, Acquisition 2, example 1. Calibrated tonometric signal in steady state. The calibrated tonometric signal follows invasive pattern, though at the end of the curve, calibrated tonometric curve tends to overestimate invasive curve. Notice the extrasystolic curve, which is nicely followed by calibrated tonometric curve.
Figure 36: Patient 3, Acquisition 2, example 2. Calibrated tonometric signal in steady state. The calibrated tonometric signal follows invasive pattern, though after a few seconds, calibrated tonometric curve tends to underestimate invasive curve.
44
4.8.2 Reliability during a fast pressure change. The next curves show a fast pressure change over a period of several seconds in the invasive pressure readings.
Figure 37: Patient 2, example 1. Calibrated tonometric signal during pressure rise. Initially, the calibrated tonometric signal follows invasive pattern adequately, but when pressure continues to rise, a manifest discrepancy emerges. (Total curve = 33 seconds)
Figure 38: Patient 3, example 1. Calibrated tonometric signal during pressure rise. Initially, the calibrated tonometric signal follows invasive pattern adequately, but after a few seconds, a discrepancy emerges. (1 st acquisition. total curve = 33 seconds)
Notice that during pressure rising of patient 2, calibrated tonometric signal underestimates invasive pressure, while in patient 3, calibrated tonometric signal overestimates invasive pressure.
45
Figure 39: Patient 3 example 2. Calibrated tonometric signal during pressure rise. Again, the calibrated tonometric signal follows invasive pattern adequately, but after a few seconds, a discrepancy emerges. (2nd Acquisition. total curve = 33 seconds)
4.8.3 Reliability during a slow pressure change. The next curves show a slow pressure change over a period of several minutes in the invasive pressure readings.
Figure 40: Patient 1 at 0 and 264 seconds. Calibrated tonometric signal after a slow pressure drop. This curve represents two acquisitions of each 16 seconds with an interval of 248 seconds of slow pressure drop.
In steady state, calibrated tonometric signal nicely follows the invasive signal. Though, after a pressure drop, the calibrated signal poorly reflects arterial pressure.
Figure 41: Patient 2 at 1468 and 1534 seconds. Calibrated tonometric signal after a slow pressure drop. This curve represents two acquisitions of each 16 seconds with an interval of 66 seconds of slow pressure drop.
46 Again, after a pressure drop over a period of 66 seconds, the calibrated tonometric signal greatly overestimates invasive pressure. While invasive pressure dropped significantly, the tonometric signal stays equal,
Figure 42: Patient 3, example 1. Calibrated tonometric signal after a slow pressure Rise. This curve represents two acquisitions of each 16 seconds with an interval of 132 seconds of slow pressure rise.
Even though calibrated pressure signal clearly detects a pressure rise, it still underestimates the invasive pressure signal. (2nd acquisition at 0 and 132 seconds).
Figure 43: Patient 3, example 2. Calibrated tonometric signal after a slow pressure drop. This curve represents two acquisitions of each 16 seconds with an interval of 66 seconds of slow pressure drop. (1 st acquisition at 132 and 198 seconds).
While initially, calibrated tonometric values tend to underestimate a stable high pressure, after a slow pressure drop of 66 seconds, the tonometric signal adequately predicts invasive pressure signals. In this case, even after 66 seconds, the tonometric signal adequately describes the invasive pressure signal.
47
Figure 44: Patient 3, example 3. Calibrated tonometric signal after a slow pressure drop. This curve represents two acquisitions of each 16 seconds with an interval of 132 seconds of slow pressure drop. (2nd acquisition at 198 and 330 seconds).
Even after a large interval, calibrated tonometric signal nicely predicts invasive pressure signal.
Figure 45: Patient 3, example 4. Calibrated tonometric signal after a slow pressure drop. This curve represents two acquisitions of each 16 seconds with an interval of 76 seconds of slow pressure drop. (2 nd acquisition, at 1541 and 1617 seconds).
Again, even after a large interval, calibrated tonometric signal nicely predicts invasive pressure signal.
Figure 46: Patient 3, example 5. Calibrated tonometric signal after a slow pressure Rise. This curve represents two acquisitions of each 16 seconds with an interval of 198 seconds of slow pressure rise. Again, even after a large interval, calibrated tonometric signal nicely predicts invasive pressure signal. (2 nd acquisition, at 1782 and 1980 seconds).
48
4.8.4 Conclusions on short-time reliability of the calibrated tonometric signal. In most cases the calibrated tonometric signal seems to follow invasive pressure signal adequately ; in the cases where a deviation occurs, there is a very high degree of unpredictability. As shown in the examples different scenarios can take place: The tonometric signal can deviate in the same or opposite direction than invasive pressure. The tonometric signal can be adequate in the first seconds after a pressure change but untruly return to its initial values after several seconds. The tonometric signal can follow pressure change adequately in the beginning of a pressure rise, but then suggest a steady state, while true pressure continues to rise. Besides needed algorithms for detecting artefacts and recalibration, the prediction of invasive pressure based on calibrated tonometric signal acquired with this equipment is not accurate enough for relying on during a delicate operative procedure.
49
4.9 Assessment of the accuracy of the calibrated signal over a total acquisition. In another approach to assess accuracy of calibrated tonometric waves, the whole acquisition was divided in segments of 20000 points, which equals at 100 Hz 200 seconds; at 300 Hz 66 seconds. In these segments, a point was chosen where a pure tonometric signal without obvious artefacts is available. At that point parameters (offset and gain) to calibrate tonometric signal based on the invasive pressure signal are assessed. Then, these parameters are applied on the whole curve of 20000 points. In order to minimize observer bias the first possible sequence for calibration was taken. Following curves are from patient 3, first acquisition, at 100HZ. Each curve covers 100 seconds. For every two curves, calibration parameters are determined at the beginning of the sequence; the light-blue line shows the timeframe where calibration parameters are determined. The red line shows the timeframe where accuracy of the calibrated tonometric signal is assessed. The next three sequences of invasive arterial (dark-blue), tonometric (purple) and calibrated tonometric (green) signals of patient 3, first acquisition at 1600, 1800 and 2000 seconds illustrate the procedure for evaluating the accuracy of the calibrated tonometric signal. Each evaluation is done over sequence of two curves. The light-blue line represents the timeframe where calibration values are determined. The red line represents the timeframe where the accuracy of the tonometric signal is determined.
Figure 47: sequence of 200 seconds. A calibration of the tonometric signal is done in the beginning. Evaluation of the accuracy of the calibrated signal is done at the end of the sequence.
50
Figure 48: Two times a sequence of 200 seconds. Notice that calibration and evaluation has to be done when invasive and tonometric signal are reliably enough. Visual assessment is done for choosing an adequate timeframe. (Light-blue and red rectangles)
51 For the whole procedures, Arterial systolic, diastolic values and Pulse Pressure, and Calibrated Tonometric systolic, diastolic values and Pulse Pressure for each investigated sequence are given in appendix 5. Also, the ratios between Arterial and calibrated tonometric values of Diastolic pressure, Systolic pressure and Pulse pressure are given in the last thee columns in percentages. These percentages are graphically represented in figure 49
Figure 49: Patient 1. Graphical representation of ratios between arterial and calibrated tonometric systolic, diastolic and pulse pressure for every frame of 66 seconds of the total acquisition. Values higher than 180% are set at 180%. Exact data can be found in appendix 5. Most of the time, values are between 80 and 120%.
A value of 110% means that invasive arterial pressure is 10% higher than calibrated tonometric pressure. When putting these values in groups in order to evaluate the percentages of total operating time that a calibrated tonometric signal (with no artefacts) is within a certain range, following results are obtained:
< 60%< 61-65% 66-70% 71-75 % 76 - 80 % 81-85 % 86-90 % 91-95 % 96-100 % 101-105 % 106-110 % 111-115 % 116-120 % 121-125% >125
syst 0 2 0 1 1 3 9 7 6 10 9 0 0 0 7
diast pulse 2 0 0 0 2 0 0 4 3 4 5 3 7 8 11 5 3 11 4 5 4 5 2 3 1 0 2 3 11 5
52
The same procedure was done for Patient 2. Manifest discrepancies between tonometric and invasive pressure values were noticed.
Figure 50: Patient 2. Graphical representation of ratios between arterial and calibrated tonometric systolic, diastolic and pulse pressure for every frame of 200 seconds of the total acquisition. Values higher than 160% are set at 160%. Absolute data can be found in appendix 5. Most of the time, values are between 80 and 120%.
Following table and graphic show in how many curves accuracy of calibrated tonometric signal is between given percentages
< 60%< 61-65% 66-70% 71-75 % 76 - 80 % 81-85 % 86-90 % 91-95 % 96-100 % 101-105 % 106-110 % 111-115 % 116-120 % 121-125% >125
syst 0 1 0 0 0 0 3 2 6 14 5 10 3 0 4
diast 0 0 0 1 1 0 2 3 10 6 7 6 4 5 3
pulse 1 0 0 2 1 2 1 3 7 6 8 3 7 0
53 The same procedure was done for Patient 3, where more manifest discrepancies between tonometric and invasive pressure values were noticed. Table with values are found in appendix 5
Figure 51: graphical representation of ratios between arterial and calibrated tonometric systolic, diastolic and pulse pressure for every frame of 200 seconds of the total first acquisition of patient 3. Most of the time, values are between 80 and 120%.
Following table and graphic show in how many curves accuracy of calibrated tonometric signal is between given percentages. syst diast pulse 125 1
54
Figure 52: graphical representation of ratios between arterial and calibrated tonometric systolic, diastolic and pulse pressure for every frame of 66 seconds of the total second acquisition of patient 3. Most of the time, values are between 80 and 120%.
Following table and graphic show in how many curves accuracy of calibrated tonometric signal is between given percentages.
< 60%< 61-65% 66-70% 71-75 % 76 - 80 % 81-85 % 86-90 % 91-95 % 96-100 % 101-105 % 106-110 % 111-115 % 116-120 % 121-125% >125
syst 0 0 0 0 2 0 2 6 9 13 3 1 1 0 1
diast pulse 0 0 0 0 0 0 1 0 0 2 3 1 2 6 6 4 10 5 10 7 3 5 1 3 0 3 1 1 1 1
Differences in pulse pressure between invasive pressure signal and calibrated tonometric pressure signal can be caused by the method of determination of diastolic and systolic pressure. If the timeframe where highest and lowest signal amplitude are determined covers multiple pulses, a swift pressure change can falsely illustrate a large pulse pressure. If the ratio between pulse pressures is determined, a falsely high or low value is obtained. This has no hemodynamic correlate and should not be considered a true inaccuracy. This type of artefact has no relevant implications on ratios of diastolic and systolic pressures.
55
From these data, one can state that for patient 3, calibrated tonometric signal after 200 seconds is within 5% of invasive arterial pressure, when only a simple calibration algorithm is used. In the other acquisitions, tonometric assessment of arterial pressure is not as adequate. This means that tonometry could prove to be useful if one can identify and correct the factors that influence signal reliability. Also needed is an algorithm that determines the quality of the signal in order only to provide pressure estimations based on signals without artefacts. When analysing the curves, human evaluation of signal quality was done in order to make calibration only on data that did not include artefacts. In order to be of clinical use, the algorithm has to be able to do so by itself. As can be seen in figure 53, an artefact in the tonometric signal can be an indicator that the tonometric position has changed slightly, which means the tonometric signal should be recalibrated; if no recalibration is done, the signal is highly unreliable. An algorithm should be written that can decide whether recalibration is required. More refined tonometric devices pretend to include such algorithms.
Figure 53: the artefact in tonometric signal clearly initiates an unreliable calibrated tonometric signal. An algorithm should recognise these events and subsequently recalibrate tonometric signal. (Patient 1 at 3630 seconds.)
These observations show that most of the time, the calibrated tonometric signal has an accuracy that could be acceptable for clinical use as an alternative for invasive pressure monitoring. Though, since a considerable error frequently arises, and its occurrence is not predictable with actual equipment, it is not accurate enough to rely on during an operative procedure.
56
5 Tonometric artefacts caused by cuff inflation As can be seen on the trend curves of patient 1 (see chapter 4.6.2), the first artefact shows a much higher peak pressure than subsequent artefacts, and there seems to be a correlation between the highest systolic pressure of the artefact and the mean blood pressure at that moment. When comparing the highest Tonometric Mean Pressure with the mean arterial pressures during cuff-inflation for the three patients, there is a clear correlation between both parameters. (For absolute values, see appendix 6)
57 When observing tonometric curves of patient 1, one can distinguish a prolonged influence after release of cuff pressure on the tonometric signal. As can be seen, this influence on the tonometric signal exists for several minutes. Most probably, this is due to physiological changes in the sensed area of the wrist as a result of cuff-inflation.
Figure 54: influence of cuff-inflation on tonometric mean pressure during stable arterial mean pressure over a period of 11 minutes.
Not only is there a manifest peak in the pressure signal, but also a consecutive rise and slight drop of tonometric pressure as long as 120 seconds after the beginning of cuff inflation or 106 seconds after cuff release. (Time between two cuff inflations is 5 minutes). When taking a close look at tonometric waveforms during cuff inflation, one can distinguish different phases of events.
Figure 55: Pattern of tonometric signal during cuff-inflation in patient 1. Description of the tonometric curve during different phases of cuff-inflation. In this curve, acquisition was done at 300 Hz. This curve consists of 10000 points, which equals 33 seconds.
A plausible explanation of this phenomenon could be the following: 1. After the cuff starts to inflate, readily, venous pressure starts to rise, while arterial pressure distally of the cuff doesn’t change a lot. Since venous stasis also causes increased pressure on the tonometer, it senses increased pressure. One can observe that total tonometric pressure rises very significantly, while pulse pressure diminishes only modestly. This can at least partially be explained by the fact that the rigid bracelet has a large contact-area with the wrist-surface, while the sensor has a small contact with
58 the wrist. So a small pressure rise against the bracelet can delivers a significant increase of force onto the small sensor. 2. Once the cuff is inflated at its maximum, it is deflated slowly, in order to detect systolic and diastolic pressure by means of oscillometry. One can see that pulse pressure drops to zero, while total tonometric pressure stagnates. This is logical since total occlusion of the artery and veins implicates that no blood can flow in or out. 3. After the cuff is inflated enough, it starts to deflate. Once the cuff is deflated to below the systolic pressure, one can see the total tonometric pressure and pulse pressure to rise quickly to a new plateau. This plateau might correspond to maximal venous stasis when venous pressure approximates arterial pressure. At this point cuff pressure is between systolic and diastolic pressure. The point with greatest pulse pressure probably corresponds with mean arterial pressure (cfr. Oscillometry principle). 4. Once the cuff pressure is further released and drops below the venous pressure, total tonometric pressure returns to its normal values remarkably fast.
When looking at figure 54, one can see that the cuff influence on the tonometric signal is still manifest after 2 minutes. Evaluation of different acquisitions of cuff inflation artefacts in the same patient show similar curves:
Figure 56: Pattern of tonometric signal during cuff-inflation in patient 1. Example 2
59 When looking at these artefacts in the other patients, one can recognise certain characteristics, though important differences merit further investigation.
Figure 57: Pattern of tonometric signal during cuff-inflation in patient 2. Example 1, low blood pressure
Figure 58: Pattern of tonometric signal during cuff-inflation in patient 2. Example 2, intermediate blood pressure
Figure 59: Pattern of tonometric signal during cuff-inflation in patient 2. Example 3, high blood pressure.
Figure 60: Pattern of tonometric signal during cuff-inflation in patient 3. Example 1
60
Figure 61: Pattern of tonometric signal during cuff-inflation in patient 3. Example 2
Figure 62: Pattern of tonometric signal during cuff-inflation in patient 3. Example 3. Note that when total pressure on the tonometric sensor is higher (as can be derived from its higher relative values in this curve), pulse pressure is much higher, although invasive pressure is roughly the same, compared to the former curve.
Figure 63: Pattern of tonometric signal during cuff-inflation in patient 3. Example 4
The characteristics of the artefact depend on different parameters. In these examples, diastolic and systolic intra-arterial pressure and pulse pressure seem to influence the pattern greatly. Possibly, physical properties of the arm, hemodynamic and vascular propeteries also have repercusions on the pattern. These observations suggest that observing these types of artefacts could be a possible source of information about vascular or other physical properties of the patient. Also, evaluating responses of the body on known influences like these cuffinflation could be used as valuable information for a more accurate, patient-depending source of calibration-parameters for the tonometric device.
61
6 Discussion & Conclusion From a technological point of view, one can state that modern electronic tonometric devices are sensitive enough to detect arterial pulsations with high enough precision. Also the hardware is capable of reproducing signals at an adequate sampling rate and with sufficient precision (cfr dampening coefficient and natural frequency). The non-invasive alternative finds its way to the clinical practice only hesitatingly, even when tonometric devices are on the market for over 20 years now. This is rather remarkable considering the advantages of non-invasive blood pressure measurement compared with invasive continuous blood pressure measurement. Which factors are responsible for the fact that the transfer from basic technology to clinical practice is so hard to take for these devices? 1. Positioning In the first place the procedure for the positioning of the sensor we used in this study is very labour-intensive, operator-dependent and in some cases almost impossible. With actual hardware, it is indispensable that not the position of the sensor, nor the direction and amount of the ‘hold down’ force are changed after the sensor has been fixed. During the clinical study it became clear that the least displacement of the sensor or fastening mechanism could make the signal of the measurement useless. Taking into account only this feature already makes the sensor less attractive. In an operative setting where after positioning, the patient is covered with sterile linen, this is an unacceptable disadvantage. For being acceptable as a clinical device it is indispensable that: 1. The localisation of the sensor on top of the artery and the adjustment of the ‘hold down’ force are considerably automated. 2. An automatic re-adjustment is possible in cases where the quality of the signal would diminish. This should be possible from a distance. In this study we tried to solve this problem by fixation of the sensor with a patch and installing a constant ‘hold down’ pressure with the bracelet. These strategies appeared a substantial improvement over manual handling of the tonometer. Nevertheless, from a technical point of view, a fully automated system (such as the Colin device) is clearly the best solution. 2. Calibration Calibration of the tonometric signal is done by an external method, usually a cuffmeasurement, based on oscillometry. Using these systems, the intrinsic restrictions of the cuff-method have their repercussions on the tonometric signal accuracy.
62 These restrictions mainly are: - Sometimes impossibility for correct measuring (obese patients); - Sometimes considerably deviating measurements (arteriosclerosis); - Difficulty or impossibility of blood pressure measurement near extreme high or low pressures. In this study, we developed and experimented with three calibration algorithms for the tonometric signal, based on the knowledge of diastolic and systolic pressure values derived from a cuff device. Our findings are: - A simple rescale algorithm based only on systolic and diastolic pressure is too limited for on-line clinical monitoring. It might be useful in certain off-line studies where calibrated tonometric data is useful for post-processing. - It might be suitable to distinguish between different events or situations and select a calibration algorithm accordingly. Which approach and algorithm suits most for calibration of the tonometric data depends on different factors: long or short datasets, manual positioning of tonometer or securely fixed, awake patient or anaesthetised, known arterial pressure change or not, scientific or clinical goal, ... - Observing the response of the body on certain influences, like cuff-inflation, could yield valuable information for calibration of the tonometric device. The analysis of such an ‘artefact’ could provide more accurate patient-depending calibrationparameters for more precise deduction of a calibrated pressure signal. In clinical practice, of course, the main objective remains to have a calibration method that allows the tonometric signal to be a sufficiently accurate approximation of an invasive blood pressure signal. Our experiments have put a light on the inherent difficulties of calibration. Our general impression is that none of the three algorithms are sufficiently intelligent to be used in real-time monitoring. From experimental point of view, but also from theoretical considerations, concerning wave propagation through peripheral arteries and the distance between the calibrating cuff (brachial) and the measuring site (radial), the second algorithm (making use of diastolic and mean cuff values of the first cycle), seemed most suitable of the three algorithms investigated. 3. Recalibration A regular automatic recalibration of the sensor is required. In actual systems, this is done by oscillometric method every three minutes. In our setting, our experiment showed that more frequent calibrations would have given more reliable results. Considering the problem of how long an oscillometry measurement remains a reliable gauger of the tonometer, it would be desirable that an algorithm of the tonometric device is able to decide whether there is too much physical interference to be still valid, and if necessary report this and recalibrate. Maybe this could be technically resolved by placing additional tonometer sensors aside the ‘arterial’ sensor, which normally should have constant load.
63 All though, it must be considered that, taking into account the margin of error of the oscillometric system, a recalibration could be denied based on the appreciation of the physician. When a severe fall of pressure makes the oscillometry measurement impossible or unreliable, it might be preferable to use an extrapolation of the last but one measurement rather than an extrapolation of the last one. It might be interesting to evaluate which pressure measurement is most reliable in case of extremely low pressure: extrapolated tonometry or new cuff- measurement. To be useful as a clinical alternative for invasive pressure measurement, the technique of arterial tonometry still has to be optimised. Suggested improvements and/or solutions formulated in this work are: automated fixation, ‘intelligent’ or at least more sophisticated calibration and recalibration algorithms. A possible approach for such an ‘intelligent’ calibration technique could be making use of a ‘transfer function’, based on comparison between segments of multiple invasive/non-invasive recordings, resulting in a ‘mean transfer function’. Although every patient has its own transfer function and inter-individual variation will probable be very significant, this approach might still be more accurate than the cuff based calibration techniques shown here. This idea will be subject of a future study of which this work represents the introductory experiments. Thinking even further, also neural networking software might be able to determine more subtle parameters and their respective weight for a more accurate estimation of invasive pressure. These kinds of approaches certainly deserve further investigation.
64
Appendix 1 : Advice from the Ethical Committee VERZOEK TOT ADVIES VAN HET ETHISCH COMITE BETREFFENDE EEN ONDERZOEKSPROJECT BIJ DE MENS 1.
TITEL VAN HET ONDERZOEK : Evaluatie van niet-invasieve bloeddrukmeting op basis van tonometrie ter hoogte van de Arteria radialis.
2.
NAAM VAN DE ONDERZOEKER(S)[tenminste één van de onderzoekers moet een persoon zijn die vast verbonden is aan de dienst (geen GSO)] : A. Kalmar, M. Struys, P. Verdonck TELEFOONNUMMER : 240 FAX : 240 49 87
32 81
UZ-DIENST OF VAKGROEP IN DE FACULTEIT GENEESKUNDE : Vakgroep Anesthesiologie UZ-DIENSTHOOFD OF VAKGROEPVOORZITTER : Prof. Dr. E. Mortier MEDEWERKERS AAN DE STUDIE : 3.
IS ER EEN FINANCIËLE SPONSOR VOOR DIT PROJECT ? Neen ZO JA (SCHRAPPEN WAT NIET PAST), FARMACEUTISCHE INDUSTRIE (WELKE ?) -
4.
5.
FWO-VLAANDEREN RUG-ONDERZOEKSFONDS ANDERE (WELKE ?)
IS HET ONDERZOEK MONOCENTRISCH ? JA - NEEN Zo ja, waar : UZ-GENT -
MULTICENTRISCH ? JA - NEEN ZO JA, WELKE CENTRA :
-
MULTINATIONAAL
IS
- NATIONAAL
HET ONDERZOEK DIAGNOSTISCH, THERAPEUTISCH, FYSIOLOGISCH, FYSIOPATHOLOGISCH, MORFOLOGISCH OF EPIDEMIOLOGISCH (SCHRAPPEN WAT NIET PAST)
65 6A.
KORTE SAMENVATTING VAN HET PROTOCOL (BINNEN DE VOORZIENE RUIMTE EN VERSTAANBAAR VOOR MENSEN NIET GESPECIALISEERD IN DE MATERIE; VERWIJS NIET ALLEEN NAAR EEN BIJGEVOEGD PROTOCOL).
Bij patiënten die een operatie ondergaan waar, om medische redenen, een invasieve bloeddrukmeting wordt verricht, worden alle fysiologische data van de anesthesiemonitor op PC overgebracht. Simultaan wordt met behulp van pentonometrie op niet-invasieve wijze de arteriële bloeddruk gemeten ter hoogte van de Arteria Radialis. Deze gegevens worden synchroon op PC opgeslagen. Nadien wordt nagegaan of de niet-invasieve drukmeting voor bepaalde indicaties een klinisch aanvaardbaar alternatief is voor invasieve A. Radialis-drukmeting.
6B.
WELKE
ZIJN DE ARGUMENTEN (THEORETISCHE, EXPERIMENTELE OF ANDERE) DIE EEN VOORDEEL LATEN VERWACHTEN VAN DE TE TESTEN NIEUWE METHODE, VAN HET TE TESTEN NIEUWE PREPARAAT, ETC. BOVEN DE GEKENDE EN REEDS GEBRUIKTE
? - Mogelijkheid tot niet-invasieve meting met bijgevolg minder risico op A. radialis trombose, digitale ischemie ten gevolge van canulatie, septisch embool, infectie. - Tijdwinst bij operatieve ingrepen met dus kortere anesthesietijd. - Breder gebruik van continue bloeddrukmeting, met mogelijk snellere interventie. 7.
WERD
EEN ANALOOG ONDERZOEK REEDS ELDERS UITGEVOERD, HETZIJ IN ZIJN GEHEEL, HETZIJ GEDEELTELIJK?
ZO JA, WAAR? WAT WAS HET RESULTAAT? WAAROM WORDT HET IN DEZE STUDIE HERNOMEN? In de literatuur is geen vermelding van synchrone metingen met dergelijke temporaire resolutie. Bovendien is er geen vergelijkende studie die specifiek de equivalentie onderzoekt tussen invasieve- en niet-invasieve Arteria Radialis drukmeting. Bijgevolg is er geen conclusie (negatief noch positief) te vinden over de klinische bruikbaarheid en indicaties als alternatief voor invasieve drukmeting.
8.
ZAL EEN CHEMISCHE SUBSTANTIE TOEGEDIEND WORDEN ? JA - NEEN LANGS WELKE WEG? NAAM EN OORSPRONG VAN DE SUBSTANTIE: AAN
WIE WORDT DE RECEPTIE, OPSLAG, VERDELING EN TERUGSTUREN VAN NIETGEBRUIKTE CHEMISCHE SUBSTANTIES TOEVERTROUWD ?
ZULLEN RADIOISOTOPEN TOEGEDIEND WORDEN ? JA - NEEN WELKE ?
66 9. INDIEN HET OM EEN NIEUWE SUBSTANTIE GAAT, HEEFT DE ONDERZOEKER KENNIS GENOMEN VAN HET VOLLEDIG TOXICOLOGISCH, DIERFARMACOLOGISCH EN HUMAAN DOSSIER? JA - NEEN ZO NEEN, LEG UIT:
10.
KEUZE VAN DE PROEFPERSONEN : GEZONDEN? JA - NEEN PATIËNTEN LIJDEND AAN: Patiënten die een chirurgische ingreep ondergaan,
Zwangere vrouwen of vrouwen die tijdens het onderzoek zwanger kunnen worden? Ja - Neen ZWANGEREN WORDEN NIET SPECIFIEK GEWEERD. AANTAL PROEFPERSONEN: 30 Leeftijd: OUDER DAN 18 JAAR GESLACHT: M/V
11A.
HEEFT
B.
MAAKT
C.
MAAKT HET EXPERIMENT DEEL UIT VAN EEN GEHEEL VAN ONDERZOEKEN WAARVAN HET DIAGNOSTISCH OF THERAPEUTISCH BELANG NIET ONMIDDELLIJK DUIDELIJK IS,
HET EXPERIMENT EEN DIAGNOSTISCH OF THERAPEUTISCH DOEL DAT ONMIDDELLIJK VOORDEEL AAN DE ONDERZOCHTE ZAL BRENGEN? JA - NEEN
HET EXPERIMENT DEEL UIT VAN EEN DIAGNOSTISCH EN THERAPEUTISCH PLAN WAARVAN MEN MAG VERWACHTEN DAT DE RESULTATEN BINNEN AFZIENBARE TIJD VOOR ANDERE ZIEKEN NUTTIG ZULLEN ZIJN? JA - NEEN
MAAR WAARVAN MAG WORDEN VERWACHT DAT DE RESULTATEN LATER TOT DIAGNOSTISCHE OF THERAPEUTISCHE TOEPASSINGEN OF TOT EEN BETERE KENNIS VAN DE FYSIOPATHOLOGISCHE MECHANISMEN ZULLEN LEIDEN ? Ja - NEEN
67 12.
WELKE
INVESTIGATIES ZIJN VOORZIEN GEDURENDE HET ONDERZOEK, HOE FREQUENT EN GEDURENDE WELKE TIJD ?
A.
ZUIVER KLINISCHE EVALUATIES, OM DE 5 minuten tot continu
B.
FUNCTIETESTS OF DYNAMISCHE PROEVEN WELKE OM DE
C.
RADIOGRAFISCHE EN/OF ISOTOPISCHE INVESTIGATIES WELKE OM DE
13.
D.
BLOEDAFNAMEN:
E.
WEEFSELAFNAME:
F.
ANDERE:
A. REKENING HOUDEND MET DE HUIDIGE GEGEVENS VAN DE WETENSCHAP, MEENT U DAT DEZE STUDIE
-
WAARSCHIJNLIJK
GEEN ENKEL RISICO INHOUDT : De studie veroorzaakt wellicht geen additioneel risico bovenop het risico eigen aan de anesthesie en chirurgie.
-
EEN MOGELIJK RISICO INHOUDT. WELK, FREQUENTIE
-
ZEER WAARSCHIJNLIJK EEN RISICO INHOUDT. WELK, FREQUENTIE :
B.
WELKE ZIJN DE MEEST VOORKOMENDE BIJWERKINGEN VAN HET PREPARAAT ONDER STUDIE ?
68 14.
INFORMATIE EN TOESTEMMING VAN DE PROEFPERSONEN *
WILSBEKWAME VOLWASSENEN
JA - NEEN
WORDT
DE TOESTEMMING VAN DE PROEFPERSONEN BEKOMEN NA EEN KLARE EN OBJECTIEVE UITEENZETTING VAN HET DOEL VAN HET ONDERZOEK?
SCHRIFTELIJK : JA - NEEN MONDELING : JA - NEEN
ZO NEEN, WAAROM NIET ?
WORDT IN DIT LAATSTE GEVAL DE TOESTEMMING GEGEVEN DOOR ANDEREN DAN DE PROEFPERSONEN ? JA - NEEN ZO JA, DOOR WIE ?
ZIJN ER SPECIALE GROEPEN : EIGEN STUDENTEN, EIGEN PERSONEEL ?
*
WILSONBEKWAME VOLWASSENEN JA - NEEN (= PSYCHIATR. PATIËNTEN, PERSONEN IN DE ONMOGELIJKHEID HUN WIL TE UITEN, ...)
WORDT DE TOESTEMMING GEGEVEN DOOR ANDEREN DAN DE PROEFPERSONEN ? JA - NEEN ZO JA, DOOR WIE ?
*
KINDEREN
JA - NEEN
WORDT
DE TOESTEMMING VERANTWOORDELIJKEN ? JA - NEEN
15.
GEVRAAGD
VAN
HUN
WETTELIJKE
IS HET INFORMATIEFORMULIER VOOR DE PROEFPERSONEN IN BIJLAGE GEVOEGD JA - NEEN ZO NEEN, WAAROM NIET ?
16.
IS HET FORMULIER VOOR SCHRIFTELIJK CONSENT IN BIJLAGE GEVOEGD ? JA - NEEN
69 17.
ZULLEN
DE PERSONEN IN DE LOOP VAN DEZE STUDIE VOORTDUREND ONDER MEDISCH TOEZICHT STAAN ? JA - NEEN
WIE IS DE TOEZICHTHOUDENDE GENEESHEER ? Alain Kalmar + staflid anesthesie ZAL DIT TOEZICHT, ZO NODIG, VERZEKERD KUNNEN WORDEN TIJDENS DE UREN DIE OP DE STUDIE VOLGEN ? JA - NEEN ALS
DE PERSOON NAAR HUIS TERUGKEERT TIJDENS DE UREN DIE OP HET ONDERZOEK VOLGEN, ZAL IN GEVAL VAN NOOD SNEL CONTACT MET EEN GENEESHEER KUNNEN OPGENOMEN WORDEN ? JA - NEEN (Niet van toepassing)
NAAM VAN DEZE GENEESHEER ? Anesthesist van wacht
18.
19.
IS DE ONDERZOEKER TEGEN EVENTUELE ONGEVALLEN VERZEKERD ? Ja A.
DOOR WELKE VERZEKERINGSPOLIS BENT U VERZEKERD ? (VERWIJZEN NAAR EEN BIJGEVOEGD DOCUMENT VOLSTAAT NIET) Verzekering van het Universitair Ziekenhuis Gent.
C.
WAT
IS DE OMVANG VERZEKERINGSMAATSCHAPPIJ?
VAN
DE
DEKKING
WORDT DEZE STUDIE DOOR DE INDUSTRIE GESPONSORD ?
DOOR
DE
JA - NEEN
JA, DIENT DE ‘HANDLING FEE’ BETAALD TE WORDEN NA ONTVANGST VAN DE FACTUUR.
ZO
IK VERKLAAR DE GEHELE VERANTWOORDELIJKHEID VAN HET HIERBOVEN VERMELD PROJECT OP MIJ TE NEMEN EN BEVESTIG DAT VOOR ZOVER DE HUIDIGE KENNIS HET TOELAAT, DE GEGEVEN INLICHTINGEN MET DE WERKELIJKHEID OVEREENSTEMMEN. DE ONDERZOEKER,
HET U.Z. DIENSTHOOFD OF DE VAKGROEPVOORZITTER (VOOR AKKOORD)
DATUM : NAAM : Dr. A. Kalmar HANDTEKENING :
DATUM : NAAM : Prof. Dr. E. Mortier HANDTEKENING :
HET UZ-DIENSTHOOFD OF DE VAKGROEPVOORZITTER BETROKKEN DIENSTEN (VOOR AKKOORD) DATUM : NAAM : HANDTEKENING :
VAN
EVENTUELE
DATUM : NAAM : HANDTEKENING :
ANDERE
70
Informatieblad voor de Patiënt ‘Vergelijkende studie invasieve/niet-invasieve continue bloeddrukmeting’
Geachte heer/mevrouw, U bent gevraagd deel te nemen aan een onderzoek waarbij we willen nagaan in welke mate een nieuwe methode van bloeddrukmeting evenwaardig is aan een actueel veel gebruikte techniek. Procedure Om tijdens een operatie de bloeddruk continu te kunnen controleren bestaat de huidige methode erin een slagader in de pols aan te prikken waarlangs de bloeddruk wordt gemeten. Dit is een frequent uitgevoerde methode die ook bij uw geplande operatie zal worden verricht. Als u instemt met het onderzoek wordt deze klassieke bloeddrukmeting uiteraard nog steeds verricht ; daarnaast wordt een extra, nieuwe meting uitgevoerd. De nieuwe methode bestaat erin een zogenaamde kleef-tonometer zachtjes tegen de polsslagader te drukken. Deze kleef-tonometer bestaat uit een schijfje van ongeveer 5 mm doorsnee die tegen de pols wordt gehouden door middel van een kleefpleister. De slagader wordt hierbij licht samengedrukt tussen een onderarmbeen en de tonometer. Hierdoor worden de drukgolven, die bij elke hartslag ontstaan, overgedragen op het toestel. Dankzij deze techniek kan bij elke hartslag de bloeddruk worden berekend. Het is de bedoeling tijdens de operatie regelmatig met deze nieuwe techniek uw bloeddruk te meten, waarbij de meetgegevens automatisch worden opgeslagen op een computer voor latere interpretatie. Het is onze hoop mede met dit onderzoek de methode van niet-invasieve bloeddrukmeting te verbeteren zodat in de toekomst in bepaalde gevallen een alternatief bestaat voor de huidige invasieve bloeddrukmeting. Risico’s De risico’s verbonden aan deze procedure zijn uiterst miniem. De operatie wordt in geen enkel opzicht gewijzigd ; het enige verschil is dat tijdens de ingreep het schijfje tegen de pols is gekleefd, zodat op geregelde tijdstippen de extra metingen kunnen worden verricht. Vrijwillige deelname Uw deelname aan dit onderzoek is volstrekt vrijwillig. Het niet deelnemen zal geen invloed hebben op de relatie tussen U en Uw dokter, noch van invloed zijn op andere toekomstige behandelingen. Vertrouwelijk Door in te stemmen met deelname aan dit onderzoek geeft U uw dokter toestemming uw medisch dossier beschikbaar te stellen voor het onderzoek. Zoals wettelijk bepaald zullen hierbij uw persoonlijke gegevens vertrouwelijk behandeld worden. Infoblad patiënt : ‘Vergelijkende studie invasieve/niet-invasieve continue bloeddrukmeting’ dd 14/12/2002, versie 1
71
Informed consent (zie informatieblad voor de patiënt dd 14/12/2002, versie 1) Ik ondergetekende, ………………………………………………., verklaar hierbij, dat ik het inlichtingsformulier (dd 14/12/2002, versie 1), met betrekking tot de ‘Vergelijkende studie invasieve/niet-invasieve continue bloeddrukmeting’ grondig gelezen heb. Bovendien werd mij op bijkomende vragen de nodige uitleg verschaft. Ik begrijp, dat mijn deelname geheel vrijwillig is en dat ik mij ten allen tijde uit de studie kan terugtrekken. Weigering om deel te nemen aan de studie zal in het geheel geen invloed hebben op de kwaliteit van de aan mij toegediende zorgen. Ik begrijp, dat alle bekomen gegevens geanonimiseerd zullen worden bij de verwerking van de data.
Gelezen en goedgekeurd,
………………….. (naam patiënt)
………………….. (naam onderzoeker)
………………….. (handtekening)
………………….. (handtekening)
…………….. (datum)
…………….. (datum)
Informed consent blad : ‘Vergelijkende studie invasieve/niet-invasieve continue bloeddrukmeting’ dd 14/12/2002, versie 1
72
Appendix 2 : Technical specifications Technical specifications of the Invasive pressure input of the Datex AS/3 monitor 74 : Sensitivity input :
Input impedance Zero drift Non-linearity gain-drift Filter Accuracy nulling Resolution calibration Nulling time
5 V/V/mmHg 5V DC max 20mA 1010 Ohms Asystole Then Asystole = Art If Art < Adiastole And Art > 5 Then Adiastole = Art tono = lees(2, rij) 'Assess for this point the tonometric value If tono > Tsystole Then Tsystole = tono If tono < Tdiastole And tono > 5 Then Tdiastole = tono Next rij Gain = (Asystole - Adiastole) / (Tsystole - Tdiastole) For rij = 2 To 3300 'For each point of the sequence : tono = lees(2, rij) 'Assess for this point the tonometric value calibrated_tono = Adiastole + (tono - Tdiastole) * Gain 'Calculate its calibrated tonometric value schrijf 3, rij, Str$(calibrated_tono) Next rij End Sub
77 Sub teken_calibratiecurve_Click() ' plot calibrated tonometric curve in a timeframe of 20.000 points, ' with calibration parameters assessed visually and input with variable ‘calibratiepunt’ Asystole = 0 'Invasive arterial systolic pressure Adiastole = 500 'Invasive arterial diastolic pressure Tsystole = 0 'Tonometric systolic pressure Tdiastole = 500 'Tonometric diastolic pressure calibratiepunt = Val(calibratiepunt.Value) zap = 0 'variable to draw reference line ' first assess invasive and tonometric diastolic and systolic pressure For rij = calibratiepunt To calibratiepunt + 500 Art = lees(1, rij) 'assess for this point the invasive value If Art > Asystole Then Asystole = Art If Art < Adiastole And Art > 5 Then Adiastole = Art tono = lees(2, rij) 'assess for this point the tonometric value If tono > Tsystole Then Tsystole = tono If tono < Tdiastole And tono > 5 Then Tdiastole = tono 'draw a reference line in order to visualize calibration timeframe : schrijf 4, rij, 80 + zap If zap = 0 Then zap = 5 Else zap = 0 Next rij Gain = (Asystole - Adiastole) / (Tsystole - Tdiastole) 'then draw the calibrated tonometric value for the whole curve For rij = 2 To 20000 tono = lees(2, rij) calibrated_tono = Adiastole + (tono - Tdiastole) * Gain schrijf 3, rij, Str$(calibrated_tono) Next rij End sub
78 Sub evalueer_tono_Click() 'evaluate accuracy of calibrated tonometric value at a timeframe beginning 'at point ‘evaluatiepunt’ with a length of 500 points (=1.6 seconds) Asystole = 0 Adiastole = 500 Tsystole = 0 Tdiastole = 500 evaluatiepunt = Val(evaluatiepunt.Value) zap = 0 For rij = evaluatiepunt To evaluatiepunt + 500 Art = lees(1, rij) 'assess for this point the invasive value If Art > Asystole Then Asystole = Art If Art < Adiastole And Art > 5 Then Adiastole = Art Caltono = lees(3, rij) 'assess for this point the tonometric value If Caltono > Tsystole Then Tsystole = Caltono If Caltono < Tdiastole And Caltono > 5 Then Tdiastole = Caltono 'draw a reference line in order to visualize evaluation timeframe : schrijf 4, rij, 80 + zap If zap = 0 Then zap = 5 Else zap = 0 Next rij 'write in a separate table evaluation values for further interpretation 'variable ‘grafiekkolom’ indicates the number of the actual of 20.000 points rij = (Val(grafiekkolom.Value) - 5) / 4 + 1 + 3 schrijf4 1, rij, Val(rij - 3) schrijf4 2, rij, Val(evaluatiepunt) schrijf4 3, rij, Val(Asystole) schrijf4 4, rij, Val(Adiastole) schrijf4 5, rij, Val(Asystole - Adiastole) schrijf4 6, rij, Val(Tsystole) schrijf4 7, rij, Val(Tdiastole) schrijf4 8, rij, Val(Tsystole - Tdiastole) 'copy columnsAD to the appropriate columns in the large table kolom = Val(grafiekkolom.Value) cl$ = "" If kolom < 27 Then cl$ = Chr$(64 + kolom) Else cl$ = Chr$(64 + Int(kolom / 26.1)) & Chr$(65 + (kolom - 1) Mod 26) End If cl$ = cl$ & ":" kolom = kolom + 3 If kolom < 27 Then cl$ = cl$ & Chr$(64 + kolom) Else cl$ = cl$ & Chr$(64 + Int(kolom / 26.1)) & Chr$(65 + (kolom - 1) Mod 26) End If cl$ = Replace(cl$, " ", "") Columns("A:D").Select : Selection.Copy Columns(cl$).Select : ActiveSheet.Paste End Sub
79
Appendix 4 : Visual evaluation of tonometric signal Tables with data of a visual assessment of relation between tonometric and invasive pressure variation. The tables on next pages show for every 66 seconds a visual assessment of the relation between tonometric and invasive signal variation. Each timeframe of 66 seconds consists of two curves of 33 seconds, which have been evaluated seperately, so for each timeframe of 66 seconds, two scores are written. Four groups have been made for each patient : 1. The invasive and tonometric signal deviate in the same direction.
2. The invasive and tonometric signal deviate in an opposite direction, or one signal diverts while the other stays steady state.
3. Both signals show are stable, the signals are in steady state.
4. Bad signal quality, no assessment of trend possible
Results of these tabels are discussed and graphically represented at chapter 4.6.
80
Table for Patient 1 Gr. 1 0Sec 66 Sec 132 Sec 198 Sec 264 Sec 330 Sec 396 Sec 462 Sec 528 Sec 594 Sec 660 Sec 726 Sec 792 Sec 858 Sec 924 Sec 990 Sec 1056 Sec 1122 Sec 1188 Sec 1254 Sec 1320 Sec 1386 Sec 1452 Sec 1518 Sec 1584 Sec 1650 Sec 1716 Sec 1782 Sec 1848Sec
Gr. 2
Gr. 3
Gr. 4
Gr. 1 1914Sec 1980Sec 2046Sec 2112Sec 2178Sec 2244Sec 2310Sec 2376Sec 2442Sec 2508Sec 2574Sec 2640Sec 2706Sec 2772Sec 2838Sec 2904Sec 2970Sec 3036Sec 3102Sec 3168Sec 3234Sec 3300Sec 3366Sec 3432Sec 3498Sec 3564Sec 3630Sec 3696Sec Sum
Gr. 2
0
5
Gr. 3
Gr. 4
108
0
81 The following table shows for every 100 seconds a visual assessment of the relation between tonometric and invasive signal variation. Each timeframe of 100 seconds consists of two curves of 50 seconds, which have been evaluated seperately, so for each timeframe of 100 seconds, two scores are written.
Table for Patient 2 Gr. 1 0 sec 200 sec 400 sec 600 sec 800 sec 1000 sec 1200 sec 1400 sec 1600 sec 1800 sec 2000 sec 2200 sec 2400 sec 2600 sec 2800 sec 3000 sec 3200 sec 3400 sec 3600 sec 3800 sec 4000 sec 4200 sec 4400 sec 4600 sec 4800 sec 5000 sec
Gr. 2
Gr. 3
Gr. 4
Gr. 1 5200 sec 5400 sec 5600 sec 5800 sec 6000 sec 6200 sec 6400 sec 6600 sec 6800 sec 7000 sec 7200 sec 7400 sec 7600 sec 7800 sec 8000 sec 8200 sec 8400 sec 8600 sec 8800 sec 9000 sec 9200 sec 9400 sec 9600 sec 9800 sec 10000 sec 10200 sec sum
Gr. 2
Gr. 3
Gr. 4
3
4
38
7
82 The following table shows for every 100 seconds a visual assessment of the relation between tonometric and invasive signal variation. Each timeframe of 100 seconds consists of two curves of 50 seconds, which have been evaluated seperately, so for each timeframe of 100 seconds, two scores are written.
Table for Patient 3, first acquisition Gr. 1 0 Sec 200 Sec 400 Sec 600 Sec 800 Sec 1000 Sec 1200 Sec 1400 Sec 1600 Sec 1800 Sec 2000 Sec 2200 Sec 2400 Sec 2600 Sec 2800 Sec 3000 Sec 3200 Sec
Gr. 2
Gr. 3
Gr. 4
Gr. 1 3400 Sec 3600 Sec 3800 Sec 4000 Sec 4200 Sec 4400 Sec 4600 Sec 4800 Sec 5000 Sec 5200 Sec 5400 Sec 5600 Sec 5800 Sec 6000 Sec 6200 Sec 6400 Sec Sum
Gr. 2
Gr. 3
Gr. 4
18
3
40
5
The following table shows for every 66 seconds a visual assessment of relation between tonometric and invasive signal variation. Each timeframe of 66 seconds consists of two curves of 33 seconds, which have been evaluated seperately, so for each timeframe of 66 seconds, two scores are written.
Table for Patient 3, second acquisition Gr. 1 0 Sec 66 Sec 132 Sec 198 Sec 264 Sec 330 Sec 396 Sec 462 Sec 528 Sec 594 Sec 660 Sec 726 Sec 792 Sec 858 Sec 924 Sec 990 Sec 1056 Sec
Gr. 2
Gr. 3
Gr. 4
Gr. 1 1122 Sec 1188 Sec 1254 Sec 1320 Sec 1386 Sec 1452 Sec 1518 Sec 1584 Sec 1650 Sec 1716 Sec 1782 Sec 1848 Sec 1914 Sec 1980 Sec 2046 Sec 2112 Sec Sum
Gr. 2
Gr. 3
Gr. 4
9
7
45
4
83
Appendix 5 : Total acquisition comparison Tables of the comparison between invasive pressure signal and calibrated tonometric signal over the total acquisition of the entire operations. The whole acquisition is devided in segments of 20000 points, which equals 200 seconds at 100 Hz; 66 seconds at 300 Hz. In these segments, a point is chosen where a pure tonometric signal without obvious artefacts is available. At that point parameters to calibrate tonometric signal onto invasive pressure signal (offset and gain) are assessed. Then, these parameters are applied on the whole curve of 20000 points. In order to minimize observer bias the first possible sequence for calibration was taken. The accuracy of the calibrated tonometric signal is assessed at the end of the evaluated sequence.
The light-blue line represents the timeframe where calibration values are determined. The red line represents the timeframe where accuracy of the tonometric signal is determined. For the whole procedure, arterial systolic, diastolic values and pulse pressure, and calibrated tonometric systolic, diastolic values and pulse pressure for each investigated sequence are shown. Also, the ratio between arterial and calibrated tonometric values of diastolic pressure, systolic pressure and pulse pressure are given in the last three columns (in percentages). (A percentage of 95% means that arterial pressure is 5% lower than calibrated tonometric pressure.)
84 Following data are from patient 1, at 300HZ. Each curve covers 66 seconds.
Table for Patient 1 Sec Time 0 66 132 198 264 330 396 462 528 594 660 726 792 858 924 990 1056 1122 1188 1254 1320 1386 1452 1518 1584 1650 1716 1782 1848 1914 1980 2046 2112 2178 2244 2310 2376 2442 2508 2574 2640 2706 2772 2838 2904 2970 3036 3102 3168 3234 3300 3366 3432 3498 3564 3630 3696
Asyst 132 125 119 105 104 94 99 98 97 99 93 101 97 94 100 97 109 100 108 108 111 114 112 116 104 110 105 102 105 97 103 104 99 97 103 99 103 105 98 104 104 105 100 103 101 103 108 93 92 84 78 69 71 65 77 86 85
mmHg Adiast 66 62 58 50 49 45 43 44 44 44 43 46 46 45 46 47 51 49 52 54 55 55 55 56 52 53 50 48 49 46 47 48 46 45 49 48 47 48 47 48 48 50 47 48 47 47 50 44 39 33 32 30 29 28 33 39 39
APP 66 63 61 54 54 49 55 54 53 55 50 54 50 48 53 49 58 51 55 53 56 59 57 60 52 57 54 53 56 50 55 55 52 52 54 51 56 57 51 55 55 55 52 55 53 56 57 48 53 50 46 38 42 37 44 46 45
CalTsyst 135 132 134 115 108 92 110 89 95 90 93 95 99 90 96 106 85 105 110 64 145 131 78 86 141 100 120 96 101 103 120 100 84 118 112 103 97 72 118 117 97 51 119 116 98 96 25 143 103 80 75 88 75 76 72 135 82
mmHg CalTdiast 70 61 72 53 49 36 47 40 41 48 35 45 48 40 36 56 33 53 60 14 83 69 35 21 76 46 57 57 26 50 50 44 29 53 52 50 51 20 54 53 40 5 53 63 49 53 5 76 51 30 21 36 33 31 35 74 38
CalTPP 65 71 62 62 59 56 63 49 53 42 58 49 51 50 59 50 51 51 49 50 61 61 42 64 65 53 63 38 75 53 69 56 54 64 59 53 45 51 64 64 57 45 66 53 48 43 19 66 51 50 54 52 41 44 36 61 43
% Asyst/Tsyst Adiast/Tdiast Apuls/Tpuls 98 94 102 95 102 89 89 81 98 91 94 87 96 100 92 102 125 88 90 91 87 110 110 110 102 107 100 110 92 131 100 123 86 106 102 110 98 96 98 104 113 96 104 128 90 92 84 98 128 155 114 95 92 100 98 87 112 169 386 106 77 66 92 87 80 97 144 157 136 135 267 94 74 68 80 110 115 108 88 88 86 106 84 139 104 188 75 94 92 94 86 94 80 104 109 98 118 159 96 82 85 81 92 94 92 96 96 96 106 92 124 146 240 112 83 87 80 89 91 86 107 120 96 206 1000 122 84 89 79 89 76 104 103 96 110 107 89 130 432 1000 300 65 58 73 89 76 104 105 110 100 104 152 85 78 83 73 95 88 102 86 90 84 107 94 122 64 53 75 104 103 105
85 Following data are from patient 2, at 100HZ. Each evaluation covers 200 seconds.
Table for Patient 2 Sec
mmHg
mmHg
%
Time Asyst Adiast APP CalTsyst CalTdiast CalTPP Asyst/Tsyst Adiast/Tdiast Apuls/Tpuls 0 163 72 91 142 58 84 115 124 108 200 146 63 82 140 59 80 104 107 103 400 199 86 112 172 76 95 116 113 118 600 218 101 117 192 126 65 114 80 180 Here, tonometric signal is of very bad quality ; no reliable calibrated signal can be obtained. 1600 122 62 60 102 50 51 120 124 118 1800 214 90 123 82 43 39 261 209 315 2000 208 91 116 182 76 105 114 120 110 2200 179 78 100 168 68 100 107 115 100 2400 160 71 88 177 76 101 90 93 87 2600 149 67 81 146 61 84 102 110 96 2800 145 66 78 131 57 74 111 116 105 3000 136 64 71 129 53 75 105 121 95 3200 122 63 58 129 58 70 95 109 83 3400 132 69 62 130 66 64 102 105 97 3600 141 73 67 124 66 58 114 111 116 3800 152 101 51 136 69 66 112 146 77 4000 148 85 62 145 96 49 102 89 127 4200 133 84 49 149 83 66 89 101 74 4400 117 78 39 131 82 48 89 95 81 4600 129 87 41 115 75 40 112 116 103 4800 121 92 29 122 82 40 99 112 73 5000 124 86 37 119 90 29 104 96 128 5200 134 90 43 113 80 32 119 113 134 5400 118 82 35 120 86 34 98 95 103 5600 120 84 36 124 85 38 97 99 95 5800 135 88 46 124 84 40 109 105 115 6000 142 89 52 135 90 45 105 99 116 6200 146 88 58 142 91 50 103 97 116 6400 146 86 59 140 86 53 104 100 111 6600 142 85 57 143 88 54 99 97 106 6800 148 85 63 142 84 58 104 101 109 7000 142 80 62 146 81 64 97 99 97 7200 90 60 30 138 81 57 65 74 53 7400 132 81 50 99 66 32 133 123 156 7600 157 87 70 123 74 49 128 118 143 7800 157 85 72 139 79 60 113 108 120 8000 163 85 77 152 81 70 107 105 110 8200 166 85 81 132 58 74 126 147 109 8400 160 84 76 140 74 65 114 114 117 8600 166 85 81 157 79 78 106 108 104 8800 165 84 81 155 78 76 106 108 107 9000 163 84 78 162 86 75 101 98 104 9200 167 83 83 170 84 85 98 99 98 9400 162 80 81 160 77 82 101 104 99 9600 156 78 78 166 82 84 94 95 93 9800 160 81 79 153 83 70 105 98 113 10000 155 80 75 135 65 70 115 123 107 10200 165 80 84 157 73 84 105 110 100
86 Following data are from patient 3, first acquisition, at 100HZ. Each evaluation covers 200 seconds.
Table for Patient 3, first acquisition Sec Time 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000 4200 4400 4600 4800 5000 5200 5400 5600 5800 6000 6200 6400 6600
Asyst 102 104 117 105 89 113 107 111 99 92 97 89 87 90 91 83 91 88 98 103 97 92 95 94 111 115 116 110 111 115 114 115 122 120
mmHg Adiast 49 50 46 43 40 52 54 59 51 41 48 42 41 42 43 39 43 42 48 48 45 42 44 44 52 52 51 47 49 50 50 51 54 55
APP 53 54 70 62 49 60 53 52 47 50 48 46 45 47 47 43 47 46 49 54 52 49 50 49 58 62 64 62 61 64 63 64 67 64
CalTsyst 100 108 103 111 97 139 117 127 95 103 98 95 81 86 94 80 96 95 96 108 99 91 100 91 105 111 119 103 110 118 114 111 109 121
MmHg CalTdiast 47 42 45 53 42 78 58 63 51 70 39 44 42 38 42 38 45 42 47 51 47 42 45 44 49 55 52 46 48 51 50 49 50 55
CalTPP 53 66 58 57 54 60 59 63 43 32 58 50 38 47 52 41 51 52 49 56 51 49 54 46 55 55 66 57 61 67 64 62 59 65
Asyst/Tsyst 102 96 114 95 92 81 91 87 104 89 99 94 107 105 97 104 95 93 102 95 98 101 95 103 106 104 97 107 101 97 100 104 112 99
% Adiast/Tdiast 104 119 102 81 95 67 93 94 100 59 123 95 98 111 102 103 96 100 102 94 96 100 98 100 106 95 98 102 102 98 100 104 108 100
Apuls/Tpuls 100 82 121 109 91 100 90 83 109 156 83 92 118 100 90 105 92 88 100 96 102 100 93 107 105 113 97 109 100 96 98 103 114 98
87 Following data are from patient 3, second acquisition, at 300HZ. Each evaluation covers 66 seconds.
Table for Patient 3, second acquisition Sec
mmHg
mmHg
%
Time
Asyst
Adiast
APP
CalTsyst
CalTdiast
CalTPP
0
120
56
63
115
55
59
Asyst/Tsyst Adiast/Tdiast Apuls/Tpuls 104
102
107
66
131
59
71
140
71
69
94
83
103
132
213
87
126
221
85
136
96
102
93
198
182
76
106
197
74
122
92
103
87
264
162
69
92
168
71
96
96
97
96
330
122
56
65
131
57
74
93
98
88
396
118
54
63
129
58
70
91
93
90
462
111
51
60
115
50
65
97
102
92
528
111
50
60
110
52
58
101
96
103
594
109
50
58
111
53
58
98
94
100
660
117
53
64
122
51
71
96
104
90
726
120
60
60
111
59
52
108
102
115
792
126
61
64
126
62
63
100
98
102
858
125
60
65
129
63
66
97
95
98
924
121
59
61
126
61
64
96
97
95
990
118
57
60
114
56
58
104
102
103
1056
119
59
60
130
64
66
92
92
91
1122
119
58
60
120
58
61
99
100
98
1188
119
59
59
109
55
54
109
107
109
1254
123
62
61
116
61
55
106
102
111
1320
128
64
64
124
65
59
103
98
108
1386
129
62
66
127
50
76
102
124
87
1452
135
67
67
131
68
62
103
99
108
1518
135
60
75
131
67
64
103
90
117
1584
110
54
56
137
65
71
80
83
79
1650
120
56
63
115
53
62
104
106
102
1716
112
53
58
107
51
55
105
104
105
1782
116
56
60
104
53
51
112
106
118
1848
124
58
65
163
77
85
76
75
76
1914
138
65
73
155
70
85
89
93
86
1980
133
63
69
131
62
68
102
102
101
2046
134
62
71
131
71
60
102
87
118
2112
147
65
82
142
80
61
104
81
134
2178
145
67
78
138
67
71
105
100
110
2244
157
77
80
121
50
71
130
154
113
2310
160
82
78
134
71
63
119
115
124
2376
176
97
78
188
107
80
94
91
98
2442
167
91
76
190
95
94
88
96
81
88
Appendix 6 : Influence of mean pressure on cuff-artefact Comparison of maximal systolic tonometric pressure during cuff inflation with the mean arterial pressure at that moment for the three patients. Maximal mean systolic tonometric signal is registrated at each cuff artefact, together with mean arterial pressure at that moment.
Patient 2
Patient 1 Mean Tonometric Arterial Peak Pressure Pressure
95 65 62 63 74 77 70 62 67 67 66 50
125 115 108 102 102 100 92 81 92 90 83 74
1 2 3 4 5 6 7 8 9 10 11 12 13
Patient 3, st 1 Acquisition
Patient 3, nd 2 Acquisition
Mean Tonometri Arterial c Peak Pressure Pressure 56 97 59 98 60 96 57 92 61 92 67 98 97 96 70 94 67 93 60 90 60 85 57 82 65 84 74 90 75 95 70 84 67 82 71 81 72 82 74 81 71 80
Mean Tonometric Arterial Peak Pressure Pressure 135 137 119 100 94 82 78 78 75 77 88 93 81 90 87 91 88 94 85 86 73 77 99 110 84 69 104 58 126 89
14 15 16 17 18 19 20 21 22 23 24 25 26 27
Mean Tonom. Arterial Peak Pressure Pressure 99 67 108 74 123 51 124 51 118 54 106 50 60 30 55 30 57 32 123 59 117 58 107 52 95 49 90 46 87 46 85 43 95 45 106 47 105 46 108 46 98 41 93 42 98 42 98 41 97 40 106 37 97 36
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
Mean Tonom. Arterial Peak Pressure Pressure 97 35 99 36 107 33 106 31 108 40 104 37 107 39 107 39 107 39 74 32 88 40 107 54 112 41 112 40 109 39 110 38 107 38 113 38 115 38 114 42 113 42 112 41 104 41 106 39 108 38 109 37
89
References 1
Block FE : What is heart rate, anyway ? J Clin Monit 10:366, 1994. Davis RF : Clinical comparison of automated auscultatory and oscillometric and catheter-transducer measurements of arterial pressure. J Clin Monit 1:114, 1985. 3 Gorback MS : Considerations in the interpretation of systemic pressure monitoring . In Lumb PD, BryanBrown CW (eds): Complications in Critical Care Medicine. Chicago, Year Book Medical Publishers, 1988, p 296. 4 Zanchetti A, Mancia G: The centenary of blood pressure measurement: A tribute to Scipione Riva-Rocci. Hypertens 14:1, 1996. 5 Shevchenko YL, Tsitlik JE: 90th anniversary of the development of Nikolai S. Korotkoff of the auscultatory method of measuring blood pressure. Circulation 94:116, 1996. 6 Goldstein S, Killip T: Comparison of direct and indirect arterial pressures in aortic regurgitation. N Engl J Med 267:1121,1962. 7 Cohn JN: Blood pressure measurement in shock: Mechanism of inaccuracy in auscultatory and palpatory methods: JAMA 199:972, 1967. 8 Ramsy M: Blood pressure monitoring: Automated oscillometric devices. J Clin Monit 7:56, 1991. 9 Iyriboz Y, Hearon CM, Edwards K: Agreement between large and small cuffs in sphygmomanometry: A qunatitative assessment. J Clin Monit 10:127, 1994. 10 Yong PG, Geddes LA: The effect of cuff pressure deflation rate on accuracy in indirect measurement of blood pressure with the auscultatory method. J Clin Monit 3:155, 1987. 11 Von Recklinghausen H: Neue wege zur blutdruckmessung. Berlin, Springer-Verlag, 1931. 12 Posey JA, Geddes LA, Williams H et al: The meaning of the point of maximum oscillations in cuff pressure in the indirect measurement of blood pressure. Part 1. Cardiovasc Res Ctr Bull 8:15, 1969. 13 Yelderman M, Ream AK: Indirect measurement of mean blood pressure in the anesthetized patient. Anesthesiology 50: 253, 1979. 14 Geddes LA: Cardiovascular Devices and Their Applications. New York ; John Wiley, 1984 15 Friesen RH, Lichtor JL: Indirect measurement of blood pressure in neonates and infants utilizing an automatic noninvasive oscillometric monitor. Anesth Analg 60:742, 1981. 16 Kimble KJ, Darnall RA, Yelderman M: An automated oscillometric technique for estimating mean arterial pressure in critically ill newborns. Anesthesiology 54:423, 1981. 17 Khan SQ, Wardlaw JM, Davenport R et al: Use of a neonatal blood pressure cuff to monitor blood pressure in the adult finger: Comparison with a standard adult arm cuff. J Clin Monit 14:233, 1998. 18 O’Brien E, Atkins W: Blood pressure measurement using oscillometric finger cuffs. Anaesthesia 50:743, 1995. 19 Kazamias TM, Gander MP, Franklin DL: Blood pressure measurement with doppler ultrasonic flowmeter. J Appl Physiol 30:585, 1971. 20 Zahed B, Sadove MS, Hatano S: Comparison of automated Doppler ultrasound and Korotkoff measurements of blood pressure. Anesth Analg 50:699, 1971. 21 Lee TK, Westenkow DR: Comparison of blood pressure measured by oscillometry from the supraorbital artery and invasively from the radial artery. J Clin Monit 14:113, 1998. 22 Weiss BM, Pasch T: Measurement of systemic arterial pressure. Curr Opin Anaesthesiol 10:459, 1997 23 Bruner JMR, Krenis LJ, Kunsman JM et al: Comparison of direct and indirect methods of measuring arterial blood pressure. Med Instrum 15:11, 1981. 24 Nystrom E, Reid KH, Bennett R et al: A comparison of two automated indirect arterial blood pressure meters: With recordings from a radial arterial catheter in anesthetized surgical patients. Anesthesiology 62:526,1985. 25 Gravlee GP, Brockschmidt JK: Accuracy of four indirect methods of blood pressure measurements, with hemodynamic correlations. J Clin Monit 6:284,1990. 26 Mark JB: Arterial blood pressure. Direct vs indirect measurement. In Mark JB (ed) : Atlas of cardiovascular monitoring. New York, Churchill Livingstone, 1998, p 81. 27 Penaz J: Photoelectric measurement of blood pressure, volume, and flow in the finger. Digest of the 10th International conference on Medical and Biological Engineering 1973, p 104. 28 TNO Biomedical Instrumentation Research Unit, Amsterdam, The Netherlands 29 Wesseling KH, De Wit B, Settels JJ, Klaver WH. On the direct registration of finger blood pressure after penaz. Funkt Biol Med 1982;1:245-250 30 Smith NT, Wesseling KH, de Wit B: Evaluation of two prototype devices producing noninvasive, pulsatile, calibrated blood pressure measurement from a finger. J Clin Monit 1:17, 1985. 2
90
31
Bos WJW, van Goudoever J, van Montfrans GA et al: Reconstruction of brachial artery pressure from noninvasive finger pressure measurements. Circulation 94:1870, 1996. 32 Gravenstein JS, Paulus DA, Feldman J et al: Tissue hypoxia distal to a Penaz finger blood pressure cuff. J Clin Monit 1:120, 1985. 33 de Jong JR, Tepaske R, Scheffer G-J et al: Noninvasive continuous blood pressure measurement: A clinical evaluation of the Cortronic APM 770. J Clin Monit 9:18, 1993. 34 Young CC, Mark JB, White W et al: Clinical evaluation of continuous noninvasive blood pressure monitoring: Accuracy and tracking capabilities. J Clin Monit 11:245, 1995. 35 Kemmotsu O, Ueda M, Otsuka H et al: Arterial tonometry for noninvasive, continuous blood pressure monitoring during anesthesia. Anesthesiology 75:333, 1991. 36 Weiss BM, Spahn DR, Rahmig H et al: Radial artery tonometry: Mod-erately accurate but unpredictable technique of continuous non-invasive arterial pressure measurement. Br J Anaesth 76:405, 1996. 37 S. Duke-Elder and J. Gloster, Physiology of the eye. The intra-ocular pressure, in: System of Ophtalmology, vol. IV, S. Duke-Elder, ed., Henry Kimpton, London, 1968. 38 Kemmotsu O, Ohno M, Takita K et al: Noninvasive, continuous blood pressure measurement by arterial tonometry during anesthesia in chil-dren. Anesthesiology 81:1162, 1994. 39 Cockings JGL, Webb RK, Klepper ID et al: Blood pressure monitoring—Applications and limitations: An analysis of 2000 incident reports. Anaesth Intensive Care 21:565, 1993. 40 Mark JB: Atlas of Cardiovascular Monitoring. New York, Churchill Livingstone, 1998. 41 Mandel MA, Dauchot PJ: Radial artery cannulation in 1000 patients: Precautions and complications. J Hand Surg 2:482, 1977. 42 Slogoff S, Keats AS, Arlund C: On the safety of radial artery cannulation. Anesthesiology 59:42, 1983. 43 Frank SM, Norris EJ, Christopherson R et al: Right- and left-arm blood pressure discrepancies in vascular surgery patients. Anesthesiology 75:457, 1991. 44 Urzua J, Sessler Dl, Meneses G et al: Thermoregulatory vasoconstriction increases the difference between femoral and radial arterial pres-sures. J Clin Monit 10:229, 1994. 45 Gravlee GP, Wong AB, Adkins TG et al: A comparison of radial, brachial, and aortic pressures after cardiopulmonary bypass. J Cardiothorac Anesth 3:20, 1989. 46 Mark JB: Direct arterial blood pressure monitoring. Normal waveforms. In Mark JB (ed): Atlas of Cardiovascular Monitoring. New York, Churchill Livingstone, 1998, p 91. 47 Slogoff S, Keats AS, Arlund C: On the safety of radial artery cannulation. Anesthesiology 59:42, 1983. 48 Singleton RJ, Webb RK, Ludbrook GL et al: Problems associated with vascular access: An analysis of 2000 incident reports. Anaesth Intens Care 21:664, 1993. 49 Garey C, Sprung J, Thomas P: Unusual complication of arterial catheterization. Anesth Analg 76:668, 1993. 50 Muralidhar K: Complication of femoral artery pressure monitoring. J Cardiothorac Vasc Anesth 12:128, 1998. 51 Horlocker TT, Bishop AT Compartment syndrome of the forearm and hand after brachial artery cannulation. Anesth Analg 81:1092, 1995. 52 Shinozaki T, Deane RS, Mazuzan JE: The dynamic responses of liquid-filled catheter systems for direct measurements of blood pressure. Anesthesiology 53:498, 1980. 53 Geddes LA: Handbook of blood pressure measurement. Clifton, NJ, Humana Press, 1991. 54 Gardner RM: Direct blood pressure measurement: Dynamic response requirements. Anesthesiology 54:227, 1981. 55 Schwid HA: Frequency response evaluation of radial artery catheter-manometer systems: Sinusoidal frequency analysis versus flush method. J Clin Monit 4:181, 1988. 56 Miller R.D. : Anesthesia Fifth edition. Churchill Livingstone, 2000. 57 Kleinman B: Understanding natural frequency and damping and how they relate to the measurement of blood pressure. J Clin Monit 5:137, 1989. 58 Hipkins SF, Ruften AJ, Runciman WB: Experimental analysis of catheter-manometer systems in vitro and in vivo. Anesthesiology 71:893, 1989. 59 Kleinman B, Powell S, Kumar P et al: The fast flush test measures the dynamic response of the entire blood pressure monitoring system. Anesthesiology 77:1215, 1992. 60 Kleinman B, Powell S, Gardner RM: Equivalence of fast flush and square wave testing of blood pressure monitoring systems. J Clin Monit 12:149, 1996. 61 Geddes LA, Bourland JD: Technical note: Estimation ol the damping coëfficiënt of fluid-filled, cathetertransducer pressure-measuring sys-tems. J Clin Engineering 13:59, 1988. 62 Gardner RM: Accuracy and reliability of disposable pressure trans-ducers coupled with modern pressure monitors. Grit Care Med 24:879, 1996. 63 McDermott M, Dearlove OR: Nasogastric contents short-circuit pressure transducer. Anaesthesia 52:719, 1997.
91
64
Gardner RM, Hollingsworth KW: Optimizing the electrocardiogram and pressure monitoring. Crit Care Med 14:651, 1986. 65 Eather KF, Peterson LH, Dripps RD: Studies of the circulation of anesthetized patients by a new method for recording arterial pressure and pressure pulse contours. Anesthesiology 10:125, 1949. 66 Abrams JH, Cerra F, Holcroft JW: Cardiopulmonary monitoring. In Wilmore DW (ed): Care of the Surgical Patient. New York, Scientific American Medicine, 1989. 67 O'Rourke MF, Yaginuma T: Wave reflections and the arterial pulse. Arch Intern Med 144:366, 1984. 68 Murgo JP, Westerhof N, Giolma JP et al: Aortic input impedance in normal man: relationship to pressure wave forms. Circulation 62:105, 1980. 69 Franklin SS, Weber MA: Measuring hypertensive cardiovascular risk: The vascular overload concept. Am Heart J 128:793, 1994. 70 Rooke GA: Systolic pressure variation as an indicator of hypovolemia. Curr Opin Anaesthesiol 8:511, 1995. 71 Perel A, Pizov R: Cardiovascular effects of mechanical ventilation. In Perel A, Stock MC (eds): Handbook of Mechanical Ventilatory Support. Baltimore, Williams & Wilkins, 1992, p 51. 72 Laeder Cath, Laboratoires pharmaceutiques Vygon 95440 Ecouen France 73 PMSET 1DT-XX Becton Dickinson Critical Care Systems Pte Ltd, 198 Yishun Avenue 7 Singapore 768926 74 DATEX/Division of Instrumentarium P.O. Box 446, SF-00101 Helsinki, Finland 75 Millar Instruments, Inc Houston, Texas 77223-0227 USA
