Download Evaluation of an ultrasonic blood volume monitor.

Nephrol Dial Transplant (1998) 13: 2098–2103
Nephrology
Dialysis
Transplantation
Technical Report
Evaluation of an ultrasonic blood volume monitor
Christian Johner1, Paul William Chamney2, Daniel Schneditz3 and Matthias Krämer1
1Fresenius Medical Care, Innovation and Technology, Bad Homburg, Germany, 2Lister Hospital, Stevenage, UK, and
3Beth Israel Medical Center, New York, USA
Abstract
Background. Hypotension complicates approximately
30% of all dialysis treatments. Although the genesis of
hypotension is multifactorial, hypovolaemia is thought
to play a major role as a direct result of decreased
blood volume, particularly during ultrafiltration. The
described blood volume monitor enables blood volume
to be measured continuously by a non-invasive
technique.
Methods. The blood volume monitor is based on the
principle that the total protein concentration, the sum
of haemoglobin and plasma proteins in the vascular
space, changes during ultrafiltration. Changes of total
protein concentration are determined from the velocity
of sound waves in blood, measured using a cuvette in
the extracorporeal circuit designed for this purpose.
The precision of the blood volume monitor was evaluated in 180 dialysis treatments in 49 patients. The
relative blood volume obtained by the monitor was
compared with a standard reference method involving
calculation of relative blood volume from serial measurements of haemoglobin.
Results. A very good agreement between the two
methods was achieved (SD=1.70%, r>0.96). The
results showed no sensitivity to changes in serum
sodium concentration (range 130–145 mmol ). The
‘noise’ introduced in the blood volume signal was low
(∏0.2%, sampling rate 10 s) allowing subtle blood
volume changes to be detected with high resolution.
In addition the device enabled the measurement of
haematocrit (Hct) and haemoglobin (Hb) to be made
since this is the largest blood component determining
total protein concentration. A comparison with the
centrifuge method revealed an accuracy of ±2.9
Hct-%, and a comparison with the photometer an
accuracy of ±0.8 g Hb/dl.
Conclusion. In summary the blood volume monitor
allows precise and reliable measurement of relative
blood volume. It provides the instrumentation essential
for feedback control of relative blood volume during
dialysis.
Correspondence and offprint requests to: Dr Christian Johner,
Fresenius Medical Care, Innovation & Technology, Daimlerstr. 15,
D-61352 Bad Homburg, Germany.
Key words: blood volume; haemodialysis; haemoglobin; instrumentation; ultrafiltration; ultrasound
Introduction
Managing optimal fluid status remains one of the most
difficult problems despite advances in dialysis [1,2]. In
the long term fluid overload is a contributing factor to
cardiac stress. In particular the development of left
ventricular hypertrophy (LVH ) is a severe risk factor
in dialysis patients. More importantly, LVH is a reversible condition which can be achieved by careful monitoring of fluid status in addition to other contributing
factors [3], since cardiac death remains the leading
cause of mortality [4]. In the short term many dialysis
treatments are complicated by the onset of hypotension
[5]. There is considerable evidence that hypovolaemia
in the vascular compartment plays an important role
in the genesis of hypotension.
Factors that interfere with blood volume homeostasis such as ultrafiltration, dry weight changes, posture changes, food ingestion, dialysate osmolarity, and
antihypertensive drugs can have a profound effect on
the changes in blood volume [6–8]. Several factors
influence the degree of blood volume reduction during
ultrafiltration, the most important is the degree of
hydration of the extracellular volume. The ability to
shift fluid from the extravascular to the vascular compartment is governed by physical factors such as
capillary permeability and oncotic forces [9,10]. A
number of sympathetic and hormonally mediated
mechanisms cause vasoconstriction which promotes
the rate of vascular refilling [11,12].
To provide more comprehensive information on fluid
shifts in addition to the long- and short-term effects of
fluid overload, objective measurements are necessary.
Absolute blood volume, the volume in litres inside the
vascular compartment, may be determined by measuring the concentration of an injected tracer or dye,
which distributes homogeneously in the vascular compartment. This method is both invasive and intermittent and is not suitable for routine application in the
monitoring and control of blood volume. During the
© 1998 European Renal Association–European Dialysis and Transplant Association
Evaluation of an ultrasonic blood volume monitor
2099
last decade there has been considerable interest in the
development non-invasive (extracorporeal ) methods
for the determination of relative blood volume (RBV ).
The RBV is the ratio of the current blood volume to
the initial blood volume at the beginning of dialysis.
RBV(t)=
c(0)
×100%
c(t)
[RBV ]=%
(1)
where c(t) is the time dependent concentration of the
blood component, e.g. cells, haemoglobin, plasma protein, total protein. All techniques rely on the principle
that these (‘solid’) blood components remain confined
to the vascular space. As plasma water is removed by
ultrafiltration the concentration of blood components
increases.
The key clinical application of blood volume monitoring is the routine feedback control of relative blood
volume [13]. Preliminary algorithms for controlling
the blood volume depending on predefined patient
specific criteria have been applied [14] demonstrating
a reduction in morbid events such as dizziness, nausea
or symptomatic hypotension.
A number of methods have been developed for
blood volume monitoring including optical [15–17],
electrical [18], and mechanical methods [19]. In the
case of optical methods for example, it is necessary to
take into consideration different absorption characteristics of light in oxygenated and deoxygenated haemoglobin in addition to different scattering properties of
blood cells which are affected by changes in osmolarity
and plasma protein concentration. The ultrasonic technique—initially introduced by Schneditz [20] has been
adopted for the measurement of the relative blood
volume for several reasons, among these is the availability of high precision transient time measurement
technology and the low costs of disposable materials,
and high achievable precision of RBV measurements.
Subjects and methods
Ultrasonic measurement of blood volume
The ultrasonic measurement technique exploits the principle
that sound speed in blood depends on total protein concentration, the sum of plasma proteins and haemoglobin [20].
Changes in sound speed can be related to changes in total
protein concentration. The relative blood volume may be
determined from protein concentration.
A polycarbonate measuring cuvette is located in the arterial
line of the extracorporeal circuit before the pump as shown
in Figure 1. Blood passes through the cuvette from bottom
to top, avoiding accumulation of air bubbles. An ultrasonic
pulse is transmitted through the cuvette containing the blood.
A silicon rubber insert ensures sound coupling to the measuring cuvette. The transit time of the pulse, which is dependent
on the speed of sound, through the blood is measured by
the BVM (Figure 1). A high-precision temperature measurement (precision <0.1°C ) is required to compensate for the
dependence of sound velocity on blood temperature [21]. An
empirical function found by Schneditz [22] is used to derive
total protein concentration TPC from sound velocity and
temperature. Using the principle of mass conservation (sim-
Fig. 1. Schematic of the BVM sensor head.
ilar to Eq. 1), the RBV can be determined from TPC as
function of time. Haematocrit and haemoglobin can be
calculated from TPC by simple linear equations assuming a
mean plasma protein concentration of 72.5 g/l at the start
of treatment. During the priming procedure when the circulating saline in the extracorporeal circuit has reached a stable
temperature, the BVM performs automatic calibration. On
detection of the presence of blood, measurement of relative
blood volume commences as soon as a sufficient stable
temperature and sound velocity is reached, typically after
1–3 min.
Additional measurements
In addition to the measurement of RBV, the BVM determines
haematocrit and haemoglobin. In this study, the reference
photometer method provided the haemoglobin measurements
for comparison with the BVM. Haematocrit was determined
via the microcentrifuge. Since some sensor systems for blood
volume are known to show artefacts caused by osmolarity
changes, both sodium and potassium were measured at
periodic intervals during the treatment. Lipids (triglycerides
and cholesterol ), plasma proteins, leukocytes, thrombocytes,
urea, creatinine, anorganic phosphorus and glucose were
measured at start and end of treatment in order to assess a
possible influence of a varying blood composition on velocity
of sound.
Reference method
To obtain discrete relative blood volume measurements,
haemoglobin (Hb) and haematocrit (Hct) are frequently
used. However, Hct is subject to variation caused by osmolality changes. Low plasma osmolality causes erythrocytes to
swell and consequently Hct is increased, even if relative
blood volume does not change.
The photometric cyanmethaemoglobin method is regarded
as the most accurate standard by which haemoglobin may
be determined [23]: 10 ml of whole blood is withdrawn into
a mini-cuvette manually with a pipette and then mixed with
a reagent of hexacyanoferrate and potassium cyanide. This
2100
C. Johner et al.
procedure lyses the red cells, eliminating light scattering
effects.
Haemoglobin was determined using the Dr Lange
Miniphotometer LP2, which has a numerical resolution of
0.1 g Hb/dl. This device contains an accurate light source at
a wavelength of 560 nm and calibration is achieved automatically. However, the overall precision of these measurements
is ultimately limited by the accuracy of manual handling.
For increased precision each blood sample was measured
three times and handled in accordance with NCCLS
(National Committee for Clinical Laboratory Standards
[23]). The presence of large amounts of lipoproteins may
elevate the apparent actual haemoglobin value, because of
turbidity. For this reason patients with these conditions
(triglycerides >500 mg/dl ) were excluded.
Haematocrit was measured from whole blood taken from
arterial lines. The blood was centrifuged for at least 5 min
at 10 000 r.p.m. using heparin-coated microcentrifuge capillaries. No further anticoagulant was used, and the haematocrit was not corrected for trapped water. Usually a standard
deviation of ±0.5 Hct-% was achieved with double readings.
Patients
Forty-nine patients (29 male, 20 female) with end-stage renal
failure and requiring intermittent haemodialysis three times
a week were selected for the study. Pregnant patients, patients
with hyperlipidaemia (triglycerides >500 mg/dl ), or patients
undergoing haemofiltration or haemodiafiltration were
excluded. The treatments lasted between 2D and 4 h, the
ultrafiltration rate, varied between 390 and 1700 ml/h.
Each patient was monitored once per week for three
dialysis sessions. Informed consent was obtained for participation in the study, in addition to approval of the ethical
committee. The study was undertaken in five centres. This
multi-centre co-operation enabled all data to be acquired
within 4 months.
Fig. 2. Correlation of relative blood volume: the data measured by
the BVM are plotted against those calculated by Hb measurements
(reference method): RBV =0.97×RBV
+2.4%.
BVM
photometer
monitor (BVM ) and the RBV determined from the
photometer. The correlation was highly significant
(r>0.96, n=882). The data correlate via RBV =
BVM
0.97×RBV
+2.4%. The mean error between
photometer
the two methods was 0.07%, standard deviation (SD)=
1.70%, demonstrating the high accuracy and reproducibility of the BVM ( Figure 3). No dependencies of
the error on the different centres or ultrafiltration rates
could be observed.
Protocol
During each treatment a total of six blood samples were
withdrawn from the extracorporeal system. Samples were
taken at the start and at the end of ultrafiltration, with
additional samples at 30, 60 and 180 min after start of
ultrafiltration. A final sample was taken 10 min after ultrafiltration to monitor blood volume rebound. Haemoglobin
was measured from each sample as described above. In order
to minimize intra-observer errors concerned with pipetting,
the haemoglobin concentration was determined from the
mean of three measurements of the same blood sample.
Haematocrit was determined from a mean of two measurements by microcentrifuge.
Statistics
Relative blood volume, haematocrit, and haemoglobin were
compared to the corresponding reference method by means
of linear regression analysis. The correlation coefficient r, the
mean deviation, and standard deviation (SD) were derived.
Results
Relative blood volume measurements
Figure 2 shows the correlation between the relative
blood volume (RBV ) measured by the blood volume
Fig. 3. Frequency distribution of the deviation in RBV measurements
between the BVM and reference method. The frequency distribution
shows the percentages of measurements revealing a deviation as
indicated on the horizontal axis. Relative blood volume measurements carried out with the BVM deviate only slightly from the
reference method (Hb – measurement with photometer): deviation=
0.07%±1.70%.
Evaluation of an ultrasonic blood volume monitor
2101
Fig. 4. Intradialytic relative blood volume by BVM and by haemoglobin measurement against time. The insert shows the low noise on the
RBV signal (<0.2%).
An example of a typical dialysis treatment is displayed in Figure 4. The inlay of this figure shows that
the noise on the RBV signal is very low (<0.2%,
sampling rate 10 s).
Haemoglobin and haematocrit measurements
Figure 5 indicates the strong correlation between the
photometer Hb measurements and the BVM, respectively (r=0.92). It is apparent from Figure 6 there are
no systematic deviations between these two methods.
The mean deviation was 0.1 g/dl, SD=0.8 g/dl.
The correlation between the Hct value derived
indirectly by the BVM and the microcentrifuge Hct
measurements was significant (r=0.88). The mean
deviation in Hct was −0.5 Hct-%, SD=2.9%.
Measurement artefacts
The data from the BVM was analysed for sensitivity
to different blood compositions. No deviations from
the reference method could be correlated with concentrations of the blood components mentioned above, as
electrolytes, solutes, lipids, glucose or proteins (r<0.3).
Nevertheless the possibility of errors cannot be
excluded in case of extreme pathological blood compositions since no such results were observed during
this study.
Fig. 5. The haemoglobin measurements by the BVM are plotted vs
the photometer measurements. The results correlate well (r=0.92).
Discussion
Symptomatic hypotension can already occur at 5–10%
blood volume reduction [24]. Many patients suffering
from hypovolaemic hypotension may be resuscitated
2102
Fig. 6. Haemoglobin measurements carried out with the BVM deviate
only slightly from the reference method (photometer): deviation=
0.1 g/dl±0.8 g/dl.
by administration of as little as 200 ml of saline and
postural changes, indicating the sensitivity of the
patient to small changes in blood volume. In an average
patient with a blood volume of approximately 5000 ml,
such a saline infusion represents only a 4% change in
blood volume. In order to resolve this change, a blood
volume sensor should provide an accuracy of 2%. The
results of this evaluation have shown that the blood
volume monitor (BVM ) has a standard deviation of
1.7%, which satisfies the measurement criteria required
for routine blood volume monitoring. Furthermore, it
is necessary to consider that the resolution of the
photometer is 0.1 g/dl in the measurement of haemoglobin. This resolution already corresponds to a 1%
uncertainty in the determination of relative blood
volume (RBV ) by the photometer. The measurement
error of the BVM adds only another 0.7% to the
observed standard deviation. A clean RBV signal with
minimal noise is necessary in order to assure a rapid
detection of sudden blood volume changes. Figure 4
demonstrates the quality of the signal which has a
noise level of <0.2%. This is a precondition for using
the BVM for automatic control of blood volume.
The RBV measurement is very precise since it is
only the relative change in the concentration of blood
constituents which is important. Any technique applied
for the monitoring of relative blood volume assumes
that the absolute mass of a particular marker (such as
haematocrit or protein) stays constant throughout the
dialysis treatment. Therefore, irrespective of the absolute value or type of a particular marker at the start
of dialysis, all markers must change in the same
proportion as ultrafiltrate is removed from the blood
during ultrafiltration. In the case of the BVM, the
initial value of the aggregate of the marker, determines
the speed of sound at the start of dialysis. Changes in
the concentration of markers during ultrafiltration
C. Johner et al.
changes the speed of sound which may be related
accurately to a change in relative blood volume.
In order to control the dose of erythropoietin in an
individual patient it is necessary to monitor haematological parameters on a regular basis introducing additional treatment costs. The BVM measures haematocrit
(Hct) with an accuracy of ±2.9% and haemoglobin
(Hb) with an accuracy of ±0.8 g Hb/dl. Whereas the
error in relative blood volume, which is derived form
total protein concentration, is small, the relative error
in Hct and Hb measurement is somewhat higher,
because the BVM cannot distinguish between intracellular proteins and plasma proteins. Since plasma
protein concentration varies between 65 and 80 g/l and
the BVM assumes a mean plasma protein concentration (p.p.c.) of 72.5 g/l, the variation in p.p.c. leads to
a slightly increased variation in Hct and Hb. In patients
whose plasma protein concentration is largely deviating
from the normal range of 65–80 g/l the error in Hct
or Hb measurement by the BVM could become unacceptably high. Such patients are rare and none was
found in the study population. For such patients a
comparison of the BVM results to laboratory analyses
should be performed to assess whether the BVM shows
a sufficient precision.
The accuracy of the BVM for measurement of Hct
or Hb does not match the precision of a laboratory
analysis, but the BVM measurement is sufficient for
reliable routine use in the dialysis setting. Some of the
error introduced between the laboratory analyser and
the BVM measurement of Hct will be due to the fact
that the BVM is not disturbed by changes in osmolality
in the observed range (265–325 mOsm/kg). A decrease
in plasma osmolality, for example due to a decrease
of sodium or plasma concentration, leads to swelling of
the erythrocytes which would not be detected by the
BVM. Therefore the BVM measures a osmolaritycorrected Hct.
Conclusion
The blood volume monitor provides a continuous,
accurate and non invasive measurement of relative
blood volume ( RBV ). The measurement is not significantly affected by changes in the blood composition,
especially changes in electrolytes and small solutes.
The low noise on the RBV signal allows a precise and
rapid detection of blood volume changes. This facilitates more objective blood volume control on an
individual basis. The BVM additionally offers the
determination of haemoglobin and haematocrit continuously with an accuracy that is acceptable for
routine haematological monitoring.
Acknowledgements. The authors wish to thank Sue Cooke, Prof. Dr
Roger Greenwood, Richard Humber (Lister Hospital Stevenage,
UK ); Thomas Fernsebner, Dr Martin Gottsmann ( Kliniken
Traunstein, Trostberg, Germany); Günther Grimm, Dr Alfred
Krause, Klaus Metzner (Fresenius Medical Care International,
Innovation and Technology); Dr H.-J. Heinicke, Lutz Herrmann,
Silke Symanc ( Krankenhaus Dresden Friedrichstadt, Germany);
Evaluation of an ultrasonic blood volume monitor
Prof. Dr Nathan W. Levin, Katja Martin (Beth Israel Medical
Center, New York, USA); Prof. Dr Helmut Mann, Dr Siegfried
Stiller, Patrick Wurth( KfH Aachen, Germany); and Christiane
Rode, Prof. Dr Volker Wizemann (Georg-Haas Dialysezentrum
Gieben, Germany).
2103
13.
14.
15.
References
1. Wizemann V, Schilling M. Dilemma of assessing volume state—
the use and limitations of a clinical score. Nephrol Dial Transplant
1995; 10: 2114–2117
2. Ifudu O. The concept of ‘dry weight’ in maintenance hemodialysis: flaws in clinical application. Int J Artif Organs 1996;
19(7): 384–386
3. Kooman JP, Leunissen KML. Cardiovascular aspects in renal
disease. Curr Opin Nephrol Hypertens 1993; 2: 791–797
4. USRDS Annual Report, 1996
5. Henderson LW. Symptomatic hypotension during dialysis.
Kidney Int 1980; 17: 571–576
6. Kouw PM, Olthof CG, Gruteke P, de Vries PMJM, Meijer JH,
Oe PL. Influence of high and low sodium dialysis on blood
volume preservation. Nephrol Dial Transplant 1991; 6: 876–880
7. Bogaard HJ, de Vries JPPM, de Vries PMJM. Assessment of
refill and hypovolaemia by continuous surveillance of blood
volume and extracellular fluid volume. Nephrol Dial Transplant
1994; 9: 1283–1287
8. Leunissen KML., Dry weight in dialysis patients. Dial J 1996
9. Landis EM, Pappenheimer JR. Exchange of substances through
the capillary walls. In: Hamilton WF, Dow P ed. Handbook of
Physiology. Washington DC, 1963; 961–1034
10. Schneditz D, Roob JM, Oswald M, Pogglitsch H, Moser M,
Kenner T. Nature and rate of vascular refilling during hemodialysis and ultrafiltration, Kidney Int 1992; 42: 1425–1433
11. Mann H, Konigs F, Heintz B, Gladziwa U, Kirsten R,
Stiller S. Vasoactive hormones during hemodialysis with intermittent ultrafiltration. ASAIO Trans 1993; 36(3): M367–369
12. Enzmann G, Bianco F, Paolini F, Panzetta G. Autonomic
16.
17.
18.
19.
20.
21.
22.
23.
24.
nervous function and blood volume monitoring during hemodialysis. Int J Artif Organs 1995; 18(9): 504–508
Santoro A, Mancini E. Clinical significance of intradialytic
blood volume monitoring. Int J Artif Organs 1997; 20(1): 1–6
Stiller S, Wirtz D, Waterbar F, Gladziwa U, Dakshinamurty K,
Mann H. Less symptomatic hypotension using blood volume
controlled ultrafiltration. ASAIO Trans 1991; 37: M139–141
Aldridge C, Wilkinson JS, Fleming SJ, Greenwood RN, Cattell
WR. Continous measurement of blood hydration during ultrafiltration using optical methods. Med Biol Eng 1987; 25: 317–323
Paolini F, Mancini E, Bosetto A, Santoro A. Haemoscan: a
dialysis machine-integrated blood volume monitor. Int J Artif
Organs 1995; 18(9): 487–494
Steuer RR, Harris DH, Weiss RL. Evaluation of a noninvasive
hematocrit monitor: a new technology. Am Clin Lab 1991;
10(6): 20–22
Maeda K, Shinzato T, Yoshida F et al. Newly developed
circulating blood volume-monitoring system and its clinical
application for measuring changes in blood volume during
hemofiltration. Artif Organs 1986; 10(6): 452–459
Greenwood RN, Aldridge C, Cattell WR. Serial blood water
estimations and in-line blood viscometry: the continuous measurements of blood volume during dialysis procedures. Clin Sci
1984; 66: 575–583
Schneditz D, Moser M, Smolle-Juttner FM, Dorp E, Pogglitsch
H, Kenner T. Methods in clinical hemorheology: the continous
measurement of arterial blood density and blood sound speed
in man. Biorheology 1989; 27: 895–902
Schneditz D, Pogglitsch H, Horina J, Binswanger U. A blood
protein monitor for the continuous measurement of blood
volume changes during hemodialysis. Kidney Int 1990; 38:
342–346
Schneditz D, Heimel H, Stabinger H. Sound speed, density and
total protein concentration of blood. J Clin Chem Clin Biochem
1989; 27: 803–806
National Committee for Clinical Laboratory Standards
(NCCLS). Reference Procedure for the Quantitative
Determination of Hemoglobin in Blood Vol 4, No. 3, 1984
Maeda K, Morita H, Shinzato T et al. Role of hypovolemia in
dialysis-induced hypotension. Artif Organs 1988; 12(2): 116–121
Received for publication: 18.8.97
Accepted in revised form: 30.3.98