Long-term pressure monitoring with arterial
applanation tonometry: a non-invasive alternative
during clinical intervention ?
Technol Health Care. 2008; 16: 183-93.
Koen S. Matthys1, Alain F. Kalmar2, Michel M.R.F. Struys2, Eric P. Mortier2, Alberto P. Avolio3, Patrick
Segers4 and Pascal R. Verdonck4
1. School of Engineering and Design, Brunel University, London, United Kingdom
2. Department of Anaesthesia, Ghent University Hospital, Gent, Belgium
3. Australian School of Advanced Medicine, Macquarie University, Sydney, Australia
4. IBiTech, Ghent University, Ghent, Belgium
Background : Arterial tonometry is a non-invasive technique for continuous
registration of arterial pressure waveforms. This study aims to assess tonometric
blood pressure recording (TBP) as an alternative for invasive long-term bedside
Methods : A prospective study was set up where patients undergoing neurosurgical
intervention were subjected to both invasive (IBP) and non-invasive (TBP) blood
pressure monitoring during the entire procedure. A single-element tonometric
pressure transducer was used to better investigate different inherent error sources of
Results : A total of 5.7 hours of combined IBP and TBP were recorded from three
patients. Although TBP performed fairly well as an alternative for IBP in steady state
scenarios and some short-term variations, it could not detect relevant long-term
pressure variations at all times.
Discussion : We discuss our findings compared to existing work and elaborate on
why physiological alterations at the site of TBP measurement might be an important
source of artifacts. We conclude that at this point arterial tonometry remains a too
unreliable technique for long-term use during a delicate operative procedure and that
physiological changes at the TBP measurement site deserve further investigation.
Manufactures of medical instrumentation should consider tonometry technology for
long-term monitoring but the associated issues as presented in this work need to be
Non-invasive continuous pressure monitoring
In previous work we reviewed the development and theoretical modelling of arterial
applanation tonometry, a technique used for non-invasive and continuous
measurement of blood pressure waveforms11. The tonometer is of particular use in
clinical research studies concerning vessel wall compliance or arterial wave traveling.
It allows for a ‘quick and easy’ recording over a few cardiac cycles of pressure
waveforms at superficial arteries and such data facilitates the calculation of
haemodynamical parameters that help characterize global vessel wall properties1,2,16.
It is remarkable that the tonometer has not been able to spread into daily diagnostic
practice even though it is a simple, non-invasive device that allows for continuous
blood pressure monitoring. Two areas immediately spring to mind where potential for
a simple probe-based pressure measurement technique such as applanation
tonometry might exist : (1) ambulatory pressure monitoring and (2) clinical bedside
Tonometry as an ambulatory blood pressure monitoring device could enhance the
information of existing home ambulatory blood pressure recorders by adding beat-to-
beat waveform information to the conventional single number output for systolic,
diastolic and mean arterial pressure (SBP, DBP, MAP). But as commercial
tonometric devices are all quite motion-sensitive, ambulatory monitoring might be a
too challenging environment for now. However, in bedside monitoring during surgical
procedures, the patient is supine and there is no direct need for ambulatory
equipment. A vast amount of arterial line (A line) recording is performed as the
arterial cannulation technique remains the accepted standard for long-term blood
pressure monitoring in anaesthesia and critical care. Most commonly, radial artery
pressure is registered because it is easy to perform and rarely associated with
complications10, 14, but despite the qualities of this technique, it remains an invasive
technique with a substantial load for the patient.
The potential of long-term non-invasive monitoring with arterial applanation tonometry
has not yet been fully explored. The ability to continuously and non-invasively monitor
blood pressure in the operating room may be advantageous in a number of
situations8,9, for example during induction or during surgical procedures where beat-
to-beat measurements are essential but no blood samples are needed. Its indications
could be expanded to any procedure where up to now only intermittent cuff
monitoring is used.
The purpose was therefore to acquire combined invasive blood pressure (IBP) and
non-invasive tonometric blood pressure (TBP) recordings during major neurosurgical
operation. Using a single-element tonometric pressure transducer, we were able to
capture unprocessed waveform data which allowed to investigate the effect of
different error-introducing aspects such as transducer fixation, positioning, calibration
and recalibration which are difficult to assess with fully automated devices where
positioning and a best signal is automatically chosen among the output of multiple
transducer elements by means of an integrated software algorithm. Identifying and
understanding the individual sources of error is imperative for developing liable
pressure monitoring. As calibration of tonometric probes is most commonly
performed with conventional sphygmomanometry and frequent recalibration is an
important issue, the influence of cuff inflation proximal to the tonometric probe is also
Non-invasive continuous pressure monitoring
Set-up and measurement protocol
Data were measured in three patients, consecutively scheduled for major
neurosurgical intervention. Written informed consent and institutional approval were
obtained. Exclusion criteria were subclavian stenosis and pre-existing radial artery
cannulation. Routine continuous monitoring in the operating room included
electrocardiography, pulse oxymetry, capnography, blood pressure recording and
rectal temperature assessment. All patients remained in supine position. Patients
were kept normothermic by a forced-air warming system.
General anaesthesia was induced by intravenous propofol 2 mg.kg-1. Cis-atracurium
0.1 mg.kg-1 was used for muscle relaxation. General anaesthesia was maintained by
continuous infusion of propofol 6 mg.kg-1.h-1, remifentanil 0.1 g.kg-1.min-1 and cis-
atracurium 0,15 mg.kg-1.h-1. Patients were intubated orally and ventilated
mechanically to achieve end-tidal CO2 and oxygen saturation within normal limits.
After anaesthesia induction, a 20-gauge 8cm PE catheter (Laeder Cath, Laboratoires
pharmaceutiques, Ecouen, France) was inserted percutaneously into the left radial
artery, 1 cm proximal to the wrist. The catheter was connected via a 150 cm long (1.5
mm internal diameter) rigid pressure tubing, filled with saline to a continuous flush
pressure-transducer system (PMSET 1DT-XX Becton Dickinson Critical Care
Systems Pte Ltd, Singapore). The system was calibrated against atmospheric
pressure. The mid-axillary line was used as the zero-reference point.
At the contralateral arm, a single-element tonometric pressure transducer (model
SSD-936, Millar Instruments, Houston, TX, USA) was placed over the right radial
artery, about 1 cm proximal to the wrist.
Figure 1 : Application Fixation of the tonometer.
Correct positioning was determined by palpation and waveform evaluation on the
anaesthesia monitor. After locating the appropriate position, the transducer was
immobilized by means of a Tegaderm patch (3M Health Care, Borken, Germany).
Hold-down pressure was then adjusted by means of a custom made bracelet with
screw until optimal waveforms with maximal amplitude were obtained. During the rest
of the procedure, these settings remained unchanged.
All monitoring equipment was connected to an S5 monitor (Datex-Ohmeda, Helsinki,
Finland). Because the Datex S5 monitoring system has no standard connection for
tonometric probes, a custom connection was made for this study. Collecting all data
via only one integrated monitoring and computing system has significant advantages
(easy synchronisation of multiple signals, maximal patient-safety with minimal use of
equipment). All data from the monitor were sampled via the Collect® Software (Datex-
Ohmeda, Helsinki, Finland) package for subsequent off-line analysis. IBP and TBP
waveforms were sampled at 100Hz. The total acquisition time for the first patient was
63 min, for the second patient 170 min and for the third patient 110 min.
It is not an aim of this study to assess a particular tonometric device. Instead the
focus lies on assessing general suitability of the tonometric technique for long-term
monitoring. As such, the analysis is tailored towards answering the question: what
can we observe clinically with the tonometric technique using the set-up described
higher and how do trends in pressure changes compare to what is observed with the
invasive standard. For it is pressure change trend behaviour that will trigger an action
from the clinical observer in the operating room in the first place. Accurate trend
agreement with IBP is therefore a minimal condition to be met by TBP.
Evaluation of normalized pressure waveforms
We investigated the trend behaviour of TBP and IBP recordings per acquisition
window of 10,000 samples, avoiding the confounding influence of TBP calibration
errors by normalizing both IBP and TBP signals to their respective first complete
heart cycle in the acquisition window. The size of the acquisition window (10,000
samples) was based on the need for a workable time-interval (100s) that is small
enough to be assessed adequately in one observation but still large enough to
capture relevant pressure trends surrounding an event such as e.g. an external cuff
inflation. Subsequently we differentiated the trends identified into four possible
categories as distinguished by an experienced observer. These four categories are
shown illustratively in Figure 2 (only a 20s time interval of a complete acquisition
window of 100s is shown). The two vital cases are agreement (case A) or
disagreement (case B) of the trends in the IBP and TBP signal: IBP and TBP can
change in the same direction (case A-AGR); IBP and TBP can deviate in an opposite
direction, or one signal deviates while the other stays in steady state (case B-
Non-invasive continuous pressure monitoring
Two further cases were identified: both signals can be in steady state (case C-SS),
and finally, poor TBP signal quality, no assessment of a trend possible (case D-
Figure 2 : Four categories of trend behaviour of the tonometric signal.
Evaluation of calibration effects
It is common practice to calibrate the tonometer with diastolic and mean pressure
values from an oscillometric cuff11. However, it has been shown that an oscillometric
recording can have a substantial error margin compared to direct invasive
recordings5,7,13. Since in this study an invasive recording was performed concurrently
with a tonometric recording, we decided to investigate the potential of applanation
tonometry in the hypothetical case of having perfect calibration values available (as
taken from IBP). This approach has also been applied in other work on tonometry4.
However, in order to investigate the effect of external interference such as an inflating
and deflating cuff on the recorded signal, we did apply an oscillometric brachial cuff
at the same arm as the tonometric pressure transducer during several (randomly
chosen) time intervals, recording systolic and diastolic blood pressure values every 3
or 5 min.
We also investigated the reliability of a calibration over time. In order to do this, we
processed the acquisition windows in pairs, with a first and subsequent window. Note
that the calibration analysis thus spans a double time interval (2 x 10,000 samples, or
200s for 100Hz) as compared to the trend analysis of the normalized signals. A set of
waveforms with no obvious artifacts over a few heart cycles was chosen and
averaged by the observer at the beginning of the first of each acquisition window
pair. Calibration offset and gain values were assessed for this averaged TBP
waveform using diastolic blood pressure (DBP) and mean blood pressure (MAP)
values from the corresponding averaged IBP waveform. These calibration
parameters were then applied on the whole acquisition window pair. Near the end of
the second window, a second set of waveforms was chosen and averaged. For this
second averaged waveform, the ratios between DBP, systolic blood pressure (SBP)
and pulse pressure (PP) values of TBP and IBP were assessed, as well as the root
mean square error (RMS) of their respective differences.
We obtained a total of 2,058,000 paired data points over a period of 343 minutes in 3
patients. Table 1 shows the evaluation of normalized TBP compared to normalized
IBP by organizing the observed trend behaviour into four different categories as
described higher (A-agree; B-Disagree; C-Steady state; D-Erroneous).
Table 1 : Qualitative evaluation of trends in TBP compared to IBP.
Total (A+B+C+D) 282
Subtotal (A+B) 41
Case A-Agree 23
Case B-Disagree 18
Case C-Steady State 229
Case D-Erroneous 12
% of Total
% of Subtotal
Non-invasive continuous pressure monitoring
From the total number (282) of acquisition windows investigated (A+B+C+D), only
4% showed poor signal quality of TBP. Further, 81% of all recordings were steady
state scenarios and in this case TBP and IBP always had a very good
correspondence. Since cases with signal variations are better markers than steady
state cases on how the TBP and IBP agree, we focused on the subtotal (the
remaining 15%) of cases in agreement or disagreement alone. For those two cases
(A+B), the percentages are almost equally divided between A and B. Thus, in relation
to the subtotal of windows in which pressure variations occurred, the TBP agreed
with IBP only in 56%.
Figure 3 : Illutstration of the evolution of TBP readings related to IBP readings.
Investigating pressure variations in further detail, it appeared in all the subjects that
small variations (e.g. caused by patient ventilation) were adequately followed by TBP
(Figure 3A). Very fast changes such as an extra-systolic event are also adequately
captured (Figure 3B & C). However, the more the pressure change is longer in time
or higher in amplitude, the more TBP monitoring tends to stay invariant to these
pressure changes (Figure 3D).
Next, calibration of TBP was performed. Note that the acquisition windows
categorized higher are now analysed in pairs as explained in the methods section.
From a total of 141 pairs, 2 had a too poor signal quality to be included in the results.
An analysis of differences between TBP and IBP (calibration precision) was
performed on the remaining 139 acquisition window pairs (with each window
consisting of 10,000 data points). Overall agreement (point-by-point) between TBP
and IBP yielded an average RMS value of 31.3 12.1 mmHg among the three
patients, while the RMS for SBP, DBP and PP values resulted in 18.3 9.3 mmHg,
12.1 4.4 mmHg and 11.0 5.6 mmHg respectively. Considering the data of all
patients together, we found an RMS of 34.3 mmHg for overall agreement and an
RMS of 20.8 mmHg for SBP, 13.2 mmHg for DBP and 12.2 mmHg for PP.
Further, the ratios of SBP, DBP and PP between TBP and IBP are represented in
Table 2 and Figure 4.
Table 2 shows the number of acquisition window pairs for which Systolic Blood
Pressure (SBP), Diastolic Blood Pressure (DBP) and Pulse Pressure (PP) from
calibrated Tonometric Blood Pressure (TBP) lie within the 5% and 10% discrepancy
intervals of the respective SBP, DBP and PP from Invasive Blood Pressure (IBP),
and this for all patients together.
Table 2 : correspondence between TBP andn IBP.
Total = 139
TBP < IBP 5%
TBP < IBP 10%
# Pairs % of Total # Pairs % of Total
SBP 54 40 88 63
DBP 43 31 77 55
PP 45 32 74 53
Non-invasive continuous pressure monitoring
Figure 4 shows the number of pairs of acquisition windows (# pairs AWin) vs. ratio
(%) of Systolic Blood Pressure, Diastolic Blood Pressure and Pulse Pressure values
from Invasive Blood Pressure (I-SBP; I-DBP; I-PP) and the respective values from
calibrated Tonometric Blood Pressure (T-SBP; T-DBP; T-PP) for all patients together.
Note that the # pairs AWin in e.g. a 5% discrepancy interval (as mentioned in Table
2) are found on these plots by adding the bars in both the 95-100% (minus 5%) and
100-105% (plus 5%) interval.
Figure 4 : Distribution of the disagreement between TBP and IBP measurement.
100-105%105-110% 110-115% 115-120%120-125%
# pairs AWin
A value of 110% means that the IBP value is 10% higher than the TBP value. From
these data it is clear that when looking at the three patients together, 38.8% of SBP
values, 30.9% of DBP values and 32.4% of PP values from calibrated TBP are within
5% of IBP. Also, 63.3% of SBP, 55.4% of DBP and 53.2% of PP values lie within
the 10% discrepancy interval. Given the fairly equal distributions in Figure 4 around
100%, there is no evidence of a consistent over- or underestimating (calibration bias)
by TBP recordings. However, considering only the very large deviations (more than
25%), we did note more cases of underestimation (IBP/TBP > 125%) than
overestimation (IBP/TBP < 75%) by TBP, and this for SBP as well as for DBP and
In summary, TBP was recorded with IBP for 5.7 hours during neurosurgery. Thus,
despite the small number of patients that did not allow for an inter-subject variability
study, we did obtain a large set of waveform data for overall comparison of IBP and
TBP. Our findings showed that calibrated TBP mirrored IBP adequately in all steady
state conditions and that minute variations of blood pressure e.g. caused by patient
ventilation or very short variations such as an extra-systolic wave were well detected.
However, these kind of variations are usually of no clinical relevance unless in
specific studies. For pressure variations longer in time and with high amplitude, there
was a rather high degree of unpredictability. In nearly half of the investigated
acquisition windows TBP either deviated in the opposite way or was only adequately
following IBP in the first seconds after a pressure change to then untruly return to its
initial regime suggesting a steady state, while IBP continued to vary.
Our set-up consisted of a single-element tonometric transducer set-up, without
automated feedback and control for position and hold-down pressure, deprived of
advanced signal conditioning and using ideal calibration values as taken from IBP.
As such stripped from possible confounding factors, our findings still indicated
inaccurate trend behaviour, suggesting that the technique of TBP monitoring is for
now unreliable or not enough understood at best to replace IBP monitoring in a
clinical setting. This mere statement alone is not an entirely new result, and two other
studies with related findings are addressed below. In analogy with our theoretical
review11, we aimed however with this practical work to add on the existing knowledge
by not merely dismissing the technique as unsuitable but rather to assess how
tonometry has evolved (or not) since past efforts, and to add (physiological) insights
into its associated problems which have not been clearly investigated before.
A decade ago Weiss and colleagues concluded from short-term recordings using an
automated multi-element array transducer (SA-250, Colin, Komaki, Japan) that
tonometry was not suitable to replace invasive monitoring during major surgical
procedures17. However, we demonstrated here with a simple (more controllable) set-
up of a single-element transducer and a custom fixation mechanism, that short-term
variations can be detected reasonably well. First, this discrepancy may well be
attributed to the evolution in transducer sensitivity over the years. Second, tonometer
positioning, signal calibration and signal display in automated devices are assessed
not by the operator but by an integrated software algorithm choosing a best output
signal among multiple elements of the transducer in the tonometric device. This is a
limitation when looking into causes of signal artifacts and variations, hence why in
this study we chose a single-element tonometric pressure transducer, put in place by
the operator with precision so that unprocessed data could be displayed and
recorded from a well-known location to facilitate the later analysis of signals and
Non-invasive continuous pressure monitoring
More recent work of Steiner and colleagues in a neuro-intensive care setting also
used a fully automated tonometry device (CBM-7000, Colin Medical, San Antonio,
TX, USA) and did perform long-term recordings (60 min)15. It was concluded that the
tonometric device was not accurate enough for replacing invasive blood pressure
(IBP) monitoring because of the significant number of inaccurate measurements and
a downward signal drift. Although the quantitative analysis for comparing TBP with
IBP was commendable, the underlying causes of the inaccuracies and drift were not
Apart from movement artifacts, positioning and fixation of which the influence on
signal quality was described in earlier work6,11, the problems are mainly caused by
the dependence of the tonometer from an external calibration device, usually a
brachial cuff-measurement based on oscillometry, of which the intrinsic restrictions
have their repercussions on the tonometric signal accuracy3. The drift noticed by
Steiner and colleagues in the time following recalibration may well be the result of
physiological changes in the limb distal to the inflated cuff.
To demonstrate this hypothesis, we applied a varying external force on the arm by
means of a brachial cuff and thus artificially induced a clear and reproducible blood
pressure change. We looked at the pattern of the uncalibrated TBP during inflation
and release of the cuff, as would occur when the calibration cuff would be positioned
at the same arm as the tonometer. By capturing the unprocessed pressure
waveforms with the single-element tonometric pressure transducer, we were able to
elaborate on the physiological changes induced by cuff inflation.
When taking a close look at TBP waveforms during cuff inflation, different events can
be distinguished that are observed repeatedly and systematically after every cuff
Figure 5 : Pattern of the TBP reading during cuff inflation and deflation.
We speculate that the induced physiological phenomenon can be explained as
follows: after the initial cuff inflation, venous pressure starts to increase, while arterial
pressure distal to the cuff doesn’t change appreciably. One can observe that MAP of
TBP increases significantly, while PP diminishes only modestly. This could be
explained by venous stasis causing the rigid hold-down bracelet to tighten slightly,
which in turn delivers a significant increase of hold-down force onto the small
pressure transducer element so that it demonstrates an increased pressure.
Once the cuff is inflated to maximum level, one can see that PP of TBP drops to zero,
while MAP of TBP stagnates. This is logical since total occlusion of the artery and
veins implicates that no blood can flow in or out. After this, the cuff is deflated slowly
in order to detect SBP and DBP by means of oscillometry. Once the cuff is deflated
below the SBP, one can see MAP and PP of the TBP signal rising quickly to a new
plateau. This plateau might correspond to maximal venous stasis when venous
pressure approximates arterial pressure. Possibly, since the cuff inflation caused an
ischaemic stimulus to the smooth muscle cells in the distal arterial vessel walls, the
vessels may dilate and increase the force onto the tonometric pressure transducer.
At this point, the cuff pressure is between SBP and DBP. Once the cuff pressure is
further released and drops below the venous pressure, the TBP returns towards its
normal values remarkably fast, which suggests indeed a predominantly intravascular
cause of the artifact. Note that although almost completely restored to the normal
values at this point, it actually takes another minute of slow but steady recovery to
really get to the initial baseline (not shown). It is unclear whether the latter would
mean that extravascular physiological disturbances play a secondary role, but it is in
any case obvious that a signal disturbance has a prolonged effect that is not
negligible compared to the conventional time span for recalibration (around 3-5
minutes for hand-held as well as automated devices).
A practical study was presented in a challenging clinical setting in which we
evaluated the proof of concept of long-term radial artery pressure waveform
monitoring via non-invasive applanation tonometry, and this as an alternative for
invasive blood pressure measurement via radial artery cannulation. We analysed
long-term recordings in three patients undergoing neurosurgery. Differently from
other work, we used a single-element tonometric pressure transducer to control and
investigate the different error-introducing aspects and we also elaborated on the
possible underlying physiological influences in the lower limb during tonometric
recording and oscillometric cuff calibration.
We conclude that, in this patient study and the tested setting, the tonometric pressure
transducer was not able to detect the relevant pressure changes at all times and a
rather high degree of unpredictability was present. As such, we believe that arterial
Non-invasive continuous pressure monitoring
tonometry still remains an essentially unreliable technique for use during a delicate
operative procedure. As we discussed, physiological alterations at the site of
tonometric measurement can be an important source of artifacts and further study of
the source of these artifacts is essential for reliable long-term non-invasive
assessment of blood pressure via applanation tonometry.
Tonometry technology represents only a small niche with respect to the blood
pressure-monitoring field. Globalisation has forced some devices of the market and
new initiatives face difficulties in taking off beyond the prototype stage12.
Incorporation in cardiovascular profiling systems (e.g. the Sphygmocor® Technology
from Atcor Medical or the CVProfilor® from Hypertension Diagnostics) is the norm.
Bedside and ambulatory pressure monitoring can present new opportunities for
manufacturers of tonometric devices, provided they tackle the persistent issues
associated with long-term recording as highlighted in this work.
1. R. Armentano, S. Graf, J. Barra, et al., Carotid wall viscosity increase is related to intima-media thickening in
hypertensive patients, Hypertension 31(1 Pt 2) (1998), 534-9.
2. A.J. Bank, D.R. Kaiser, S. Rajala, et al., In vivo human brachial artery elastic mechanics: effects of smooth
muscle relaxation, Circulation 100 (1999), 41-7.
3. A. Bur, H. Herckner, M. Vlcek, et al., Factors influencing the accuracy of oscillometric blood pressure
measurement in critically ill patients, Crit Care Med 31(3) (2003), 793-9.
4. C-H. Chen, E. Nevo, B. Fetics, et al., Estimation of central aortic pressure waveform by mathematical
transformation of radial tonometry pressure. Validation of generalized transfer function, Circulation 95 (1997),
5. R.F. Davis, Clinical comparison of automated auscultatory and oscillometric and catheter-transducer
measurements of arterial pressure, J Clin Monit 1(2) (1985), 114-19.
6. G.M. Drzewiecki, J. Melbin, A. Noordergraaf, Arterial tonometry: review and analysis, J Biomech 1983,141-52.
7. G. Gravlee, S. Bauer, M. O'Rourke, et al., A comparison of brachial, femoral and aortic intraarterial pressures
before and after cardiopulmonary bypass, Anaesth. Intensive Care 17 (1989), 305-11.
8. O. Kemmotsu, M. Ueda, H. Otsuka, et al., Blood pressure measurement by arterial tonometry in controlled
hypotension, Anesth Analg 73(1) (1991), 54-8.
9. O. Kemmotsu, M. Ueda, H. Otsuka, et al., Arterial tonometry for noninvasive, continuous blood pressure
monitoring during anesthesia, Anesthesiology 75(2) (1991), 333-40.
10. M. Mandel and P. Dauchot, Radial artery cannulation in 1000 patients: precautions and complications, J Hand
Surg 2(6) (1977), 482-5.
11. K. Matthys and P.Verdonck, Development and modelling of arterial applanation tonometry: a review,
Technology and Health Care 10(1) (2002), 65-76.
12. K.G. Ng, C.M. Ting, J.H. Yeo, et al., Progress on the development of the Mediwatch ambulatory blood
pressure monitor and related devices, Blood Pressure Monitoring 9 (2004), 149-65.
13. E. Nystrom, K.H. Reid, R. Bennett, 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(4)
14. S. Slogoff, A. Keats and C. Arlund, On the safety of radial artery cannulation, Anesthesiology 59 (1983), 42-7.
15. L.A. Steiner, A.J. Johnston, R. Salvador, et al., Validation of a tonometric noninvasive arterial blood pressure
monitor in the intensive care setting, Anaesthesia 58 (2003), 448-54.
16. H. Tanaka, F.A. Dinenno, K.D. Monahan, et al., Aging, habitual exercise, and dynamic arterial compliance,
Circulation 12(102(11)) (2000), 1270-5.
17. B. Weiss, D. Spahn, H. Rahmig, et al., Radial artery tonometry: moderately accurate but unpredictable
technique of continuous non-invasive arterial pressure measurement, Br J Anaesth 76(3)(1996), 405-11.