British Journal of Anaesthesia 1998; 80: 521–524
EQUIPMENT
Error in measurement of oxygen and carbon dioxide concentrations by
the DeltatracII metabolic monitor in the presence of desflurane
T. W. L. SCHEEREN, M. KROSSA, P. MERILÄINEN AND J. O. ARNDT
Summary
In experiments in dogs on the metabolic effects
of inhalation anaesthetics, we noticed that in the
presence of desflurane, oxygen uptake (VO2)
measured with the DeltatracII metabolic monitor
seemingly increased whereas it decreased when
determined independently by the Fick principle.
This difference remained even after correction
for changes in gas concentration on addition of
an inhalation anaesthetic. Therefore, we suspected that desflurane interferes with the measurement of gas concentrations. Using different
precision gases, we found that desflurane
disturbed both the paramagnetic oxygen sensor
and the infrared carbon dioxide detector so that
the measured oxygen (when ,2 was 0.21) and
carbon dioxide concentrations were greater than
expected. These errors multiply in the computing process of oxygen uptake by the DeltatracII.
When the DeltatracII is to be used during inhalation anaesthesia, its results should be corrected
for the presence of an anaesthetic gas. More
importantly, corrections must also be made for
measurement errors of the oxygen and carbon
dioxide sensors, unless the device has been
equipped with a modified (nickel membrane)
oxygen sensor insensitive to the presence of
volatile agents. (Br. J. Anaesth. 1998; 80:
521–524)
Keywords: anaesthetics volatile, desflurane; monitoring,
oxygen; measurement techniques, calorimetry; equipment,
gas analysers
We would like to draw attention to a source of error
in the measurement of oxygen and carbon dioxide
concentrations in the presence of inhalation anaesthetics in one particular analyser. The problem arose
in experiments on the metabolic effects of desflurane
in dogs. Using the DeltatracII metabolic monitor
(Datex Instrumentation Corp., Helsinki, Finland) we
noticed that both oxygen uptake (V!O2 ) and carbon
dioxide production (V!CO 2) seemingly increased with
increasing desflurane concentration whereas it decreased with other inhalation anaesthetics. This overt
discrepancy was also surprising because cardiac output markedly decreased in association with only a
slight increase in the arteriovenous oxygen content
difference, regardless of the anaesthetic used, that is
even in the presence of desflurane. To resolve these
contradictory observations, we determined V!O2 by
the Fick principle (from the arteriovenous oxygen
content difference and cardiac output), independently of the DeltatracII.
We found that these V!O2 values were markedly
lower than those measured by the DeltatracII, even
after correction of the latter for dilution of the respiratory gases on addition of desflurane.1 Hence the
large concentrations of desflurane (MAC 6–7%)
used for anaesthesia2 and the subsequent greater
inspiratory–expiratory desflurane difference alone
did not entirely explain this considerable discrepancy. Therefore, we suspected that desflurane was
causing an error in the measurement of gas concentrations, most likely by interfering with either the
paramagnetic oxygen sensor (OM-101, Datex) or the
infrared absorption carbon dioxide detector (CX104, Datex), or both, used in the DeltatracII. This
led us to evaluate the effects of desflurane on the
measurement behaviour of both sensors in the presence of precision gases containing different concentrations of oxygen, carbon dioxide, or both.
Materials and methods
The DeltatracII is an open circuit indirect calorimeter using a gas dilution principle for determining
respiratory gas volumes required for gas exchange
measurements.3 The paramagnetic oxygen sensor
measures the difference in oxygen concentrations
between inlet I (alternating inspired gas or room air)
and inlet II (mixed expired gas) (fig. 1A). (Note that
the microphone membrane of the differential oxygen
sensor is made of plastic and covered with a metal
layer (gold) at inlet I.) Unlike in standard gas analysers, in which the plastic surface of the sensor membrane (at inlet II) is exposed exclusively to air (fig.
1C), volatile anaesthetics may reach both sides of the
sensor membrane in the DeltatracII (fig. 1A, B) and
thus affect its sensitivity.
Before each experiment, the gas sensors were calibrated with a high precision gas (5.00<0.03% carbon
dioxide–balance oxygen; Messer Griesheim, Duisburg, Germany) and the DeltatracII measurement of
THOMAS W. L. SCHEEREN*, MD, MANFRED KROSSA, JOACHIM
O. ARNDT, MD, Department of Experimental Anaesthesiology,
Heinrich-Heine-University,
Düsseldorf,
Germany.
PEKKA
MERILÄINEN, PHD, Datex-Engstrom Division, Instrumentarium
Corp., Helsinki, Finland. Accepted for publication: December 17,
1997.
*Address for correspondence: Institut für Experimentelle
Anaesthesiologie,
Heinrich-Heine-Universität,
Düsseldorf,
Moorenstr. 5, Geb. 23.02, D-40225 Düsseldorf, Germany.
522
British Journal of Anaesthesia
Figure 1 Schematic representation showing the oxygen
measurement configuration in the DeltatracII metabolic monitor
during conventional use (A) and during our bench experiments
(B) compared with the configuration in standard gas analysers
(C).
oxygen uptake was calibrated by burning ethanol
5 ml (99.9%, alcohol burning test kit of Datex).
In our bench experiments (fig. 1B), three precision
gases (Messer Griesheim) containing various oxygen
concentrations (20.9%, 30.3% or 50.1% oxygen–
balance nitrogen) were sucked (flow : 100–150 ml
min91) through the inspiratory port of the DeltatracII
(inlet I) to the metallized side of the sensor
membrane. The same gases, after equilibration with
0, 7 and 14 vol.% desflurane (vaporizer Tec 6, Ohmeda, Steeton, UK), were fed to the mixing chamber
of the DeltatracII (at a flow of 2 litre min91) from
where samples were sucked to the plastic side of the
sensor membrane (via inlet II). In this way it was
possible to recognize if the presence of desflurane at
the plastic side of the oxygen sensor membrane (via
inlet II) would disturb the oxygen concentration
measurement of the inspiratory gas sample (inlet I).
Desflurane concentrations were monitored continuously by infrared spectroscopy (Capnomac Ultima
SV multi-gas analyser, Datex).
Results
With the sensor membrane exposed to desflurane via
inlet II, the DeltatracII measured erroneously large
oxygen concentrations of the inspiratory gas sample.
This is shown by the relations between oxygen
concentrations in the precision gases and those measured by the DeltatracII in figure 2A. As expected, in
the absence of desflurane, the data points fell almost
on the line of identity, indicating that the DeltatracII
measured oxygen concentrations correctly in these
circumstances. However, in the presence of desflu-
Figure 2 Effects of desflurane on measurements of oxygen
concentrations. In A, measured oxygen concentrations deviated
from those in the precision gases at larger oxygen and desflurane
concentrations. Note that oxygen concentrations are measured
against room air so that the difference becomes 0 when air is used
as a carrier gas. The inset shows that the slopes of the resulting
curves can be related linearly to desflurane concentrations, which
alleviates correction. In B, the DeltatracII readings of ⌬O2 (i.e.
oxygen concentration difference over the sensor membrane, inlet I
vs II) were greater (19% on average) than expected in the
presence of desflurane. The broken lines indicate identity.
rane at inlet II, the DeltatracII measured the oxygen
differentials (i.e. O2 sample gas9O2 Air) correctly only with
room air as a carrier gas but gave higher readings
when the oxygen concentrations increased at inlet I.
As a result, the relationships between measured and
expected oxygen differentials retained their linearity
but were tilted anti-clockwise (fig. 2A). While the
intercepts of these relationships were negligible (0.2
for each curve), their slopes increased from 0.97 to
1.07 and 1.23 (at 0, 7 and 14 vol.% desflurane), and
could be related, by first approximation, to the
fractions of desflurane (FDesfl) (see inset in fig. 2A):
slope : 1;1.9FDesfl.
Hence, it was possible to correct the oxygen measurements for this error:
Desflurane disturbs oxygen and carbon dioxide sensors
Figure 3 Effects of desflurane on measurements of carbon
dioxide concentrations. The carbon dioxide gain error (% change
in the ratio of measured and calculated carbon dioxide
concentrations) and the desflurane fractions at inlet II (FEDesflurane)
correlated linearly.
FO2 Deltatrac − 0.209
FO2* =
1 + 1.9 × F Desfl.
+ 0.209
(1)
where *indicates a corrected value and gas concentrations are given as fractions (F).
How this measurement error affects the inspiratory
and expiratory oxygen concentration difference
across the sensor membrane (i.e. ⌬O2 between inlet I
and II) computed by the DeltatracII is shown in figure 2B. For the ranges of desflurane and oxygen concentrations studied, the data (⌬O2, as measured by
the DeltatracII:⌬O2 Deltatras against ⌬O2, as calculated
from dilution of the precision gases on addition of
desflurane:⌬O2 calc.) fitted well a linear equation (see
insert) with a slope notably steeper than the line of
identity. Thus the DeltatracII readings were always
greater (19% on average) than expected. This
constant percentage error renders the correction of
⌬O2 Deltatras simple:
∆FO2* =
∆FO2 Deltatrac
(2)
1.19
Although of less importance, desflurane also
disturbed the infrared carbon dioxide sensor. On
addition of desflurane (5, 10, and 15 vol.%),
measured carbon dioxide concentrations were 4.2,
7.8 and 10.4% greater than those calculated using a
precision gas containing 5.00 < 0.03% carbon
dioxide–19% oxygen–76% nitrogen. Thus the gain
error (% change in the ratio of measured and calculated values) increased linearly with desflurane
fractions (fig. 3) and can be compensated for by the
following equation:
FE CO2* =
FE
CO 2 Deltatrac
1 + (70 × FE Desfl. + 0.4) / 100
(3)
9FE *)V! E
o2
where FO2, FN2, FCO2 and FAn (anaesthetics) are the
respective inspiratory (FI) and expiratory (FE) gas
fractions, VE is the expired minute ventilation under
STPD conditions, and expressions marked by * indicate mean corrected values according to equations
(1)–(3). Those terms which are not directly available
at the data output string of the DeltatracII can be
calculated according to FE O2 = FI O2 *− ∆ FO2* and
FI N2 = FN2×(1 − FI An)≅FEN2 during steady state conditions.
In fact, as exemplified in table 1 for one dog, V!O2
values calculated by equation (4) agreed well with
those obtained independently by the Fick principle
(last column). Note also that in the presence of desflurane, the DeltatracII readings (first column), even
when corrected for the presence of anaesthetic gases
(second column), yielded V!O2 values which were up
to 1.7 and 1.3 times greater than the reference
method (Fick, last column). Similar results were
obtained in 10 experiments on six dogs (data not
shown).
As the V!O 2 measurements by the Fick principle
were used as a reference, we took care to measure as
accurately as possible both the arteriovenous oxygen
content difference and cardiac output. Arterial
(carotid artery) and mixed venous blood (pulmonary
artery) were sampled simultaneously in air-free
syringes and the total oxygen content in blood measured immediately in duplicate (Lex O2 Con, Lexington Instr., Waltham, MA, USA).
Cardiac output (i.e. pulmonary blood flow) was
measured continuously (beat by beat) with ultrasonic
transit-time flow probes (Transonic, Ithaka, NY,
USA) chronically implanted around the pulmonary
artery (after approval by the local District Government) and averaged over the entire blood sampling
period.
For every measurement, we calculated the worst
case error according to the law of error propagation:
the accuracy of the oxygen content analyser is <1%
for a single measurement and <0.75% for the mean
of two measurements. Hence the 95% confidence
interval for an arterial or mixed-venous sample is
within <1.5% (<2 SD) of the measured value and the
error of the calculated arteriovenous oxygen content
difference sums up to 6% (23%).
The accuracy of the ultrasound transit-time
flowmeter is <7.5% for a single stroke volume but
improves to <1.2% (7.5 √40) by averaging over the
sample period (i.e. approximately 30 s or 40 beat
min91). The resultant maximum errors of the V!O2 values determined by the Fick principle (the product of
Table 1 Effects of desflurane on measurements of oxygen (O2)
uptake. O2 uptake is shown as measured by the DeltatracII before
and after corrections and as the reference values determined by
the Fick principle. Data from one representative experiment in a
dog
O2 uptake (ml kg91 min91)
where FE:expiratory gas fractions.
Finally, with the corrected gas concentrations, V!O2
can be calculated using a modification of the
Haldane transformation for anaesthesia, as suggested
by Aukburg and colleagues1:
FE *;FIN2;FECO2 *;FE An
V!O2*:( FIo2* o2
FIo2*; FIN 2;FI An
523
(4)
Corrected
Original
according to
Desflurane DeltatracII Aukburg and
(vol. %)
readings
colleagues1
Corrected
according
to eqn
(4)
Determined
by the
Fick
principle
0
7
14
4.3
3.6
3.2
4.3
3.6
3.1
4.3
4.6
5.1
4.3
4.1
4.0
524
arteriovenous oxygen content difference and cardiac
output) add to 9 and 10% (at 7 and 14 vol.% desflurane), which is acceptable in view of the 28–65%
deviation of V!O 2 values obtained by the DeltatracII.
Of note, lung oxygen consumption, which is not
detected by the Fick measurement but is detected
with the DeltatracII, is less than 5% of total V!O 2 and
thus negligible in healthy dogs and humans.4 5
Discussion
Desflurane disturbed, in a concentration-dependent
manner, measurements of oxygen and carbon
dioxide concentrations in the DeltatracII metabolic
monitor. As a consequence, the oxygen concentration differences across the sensor membrane (inspiratory to expiratory ⌬O2) were overestimated by
approximately 20% in the presence of desflurane. As
⌬O2 is a dominant factor in the computation of V!O 2
(VO2:⌬O2VE), erroneous increases in ⌬O2 must
also yield erroneous increases in V!O 2 . In fact, the
DeltatracII yielded V!O 2 values which were up to 1.7
times greater than expected (Fick principle) (table 1)
if no corrections were applied. Hence the DeltatracII
readings must be corrected for both the addition of
desflurane (which alters respiratory gas concentrations) and also for the error in the measurement of
gas concentrations. The correction for the first problem according to Aukburg and colleagues1 yielded
V!O2 values that were up to 30% greater than the Fick
values. This discrepancy reflects the error in the
measurement of gas concentrations because it was
eliminated by the appropriate corrections using
equations (1)–(4), as evidenced by the agreement of
the corresponding V!O 2 values with the reference (see
last two columns in table 1).
The measurement errors of the gas sensors are not
unique to desflurane, as we have found that other
inhalation anaesthetics disturbed the oxygen and
carbon dioxide detectors in a similar manner (data
not shown). Yet because of the smaller concentrations used for anaesthesia, impairment of gas
concentration measurements matters little for the
computation of V!O 2.
Interference of volatile anaesthetics with carbon
dioxide detectors has been reported previously,6 the
underlying mechanism being deformation of the
infrared carbon dioxide absorption peak by collision
of carbon dioxide with other gas molecules.7
However, this carbon dioxide measurement error is
of minor relevance for calculation of V!O 2 as its
correction alone decreased the V!O 2 readings of the
DeltatracII by only 1–2%.
We have shown for the first time that volatile
anaesthetics (which are non-paramagnetic molecules) also disturb the paramagnetic oxygen sensor.
The culprit for this error seems to be the plastic surface which alters the sensitivity of the microphone
membrane of the differential oxygen sensor either by
deformation of its thickness (swelling) or change of
its dielectric constant as desflurane is absorbed.
Consistent with this view, replacing the plastic membrane by one made from nickel almost completely
eliminated the oxygen measurement error induced
by desflurane (data not shown). Standard gas monitors which use the same gas sensors (e.g. the
British Journal of Anaesthesia
Capnomac Ultima, Datex) were not affected by desflurane because the plastic surface of its microphone
membrane is always exposed to room air thus
preventing contact with volatile agents (fig. 1C).
Of note, the DeltatracII metabolic monitor has not
been designed for metabolic studies during inhalation anaesthesia. Nevertheless, it can be used for this
purpose when the appropriate corrections are made
for the changes in respiratory gas concentrations by
additional gases and, in particular, for the effects of
inhalation anaesthetics on the measurement of gas
concentrations, as we have shown. Another option
would be to use gas sensors which are not disturbed
by volatile anaesthetics. Such modified sensors will
be available soon as spare parts for the DeltatracII
metabolic monitor. The new version of the paramagnetic oxygen sensor of the Datex-Engstrom,8 as
implemented in the compact gas exchange module
M-COVX of the Datex AS/3 monitor system, is
equipped with a nickel film microphone membrane
and therefore is immune to volatile anaesthetic
agents.
It should be noted that the DeltatracII cannot be
used during anaesthesia with nitrous oxide instead of
nitrogen. In this case, the Haldane transformation
used to derive inspired from expired tidal volumes
does not apply because of the different uptake kinetics of nitrous oxide. Furthermore, it should be
remembered that our corrections apply to steady
state conditions rather than to rapidly changing gas
concentrations, and that different gas sensors may
require an individual check of its correction factors.
Finally, it is a pleasing result that oxygen uptake,
when determined correctly, does not increase, but
rather decreases in the presence of desflurane, in
common with other inhalation anaesthetics.9
Acknowledgements
We thank Hoyer Engström (Bremen, Germany) for providing a
Capnomac Ultima SV monitor (Datex) for measuring desflurane
concentrations and Mrs B. Berke for her excellent technical assistance in the experiments on dogs.
References
1. Aukburg SJ, Geer RT, Wollman H, Neufeld GR. Errors in
measurement of oxygen uptake due to anesthetic gases.
Anesthesiology 1985; 62: 54–59.
2. Eger EI II. New inhaled anesthetics. Anesthesiology 1994; 80:
906–922.
3. Meriläinen PT. Metabolic monitor. International Journal of
Clinical Monitoring and Computing 1987; 4: 167–177.
4. Light RB. Intrapulmonary oxygen consumption in experimental pneumococcal pneumonia. Journal of Applied Physiology 1988; 64: 2490–2495.
5. Loer SA, Scheeren TWL, Tarnow J. How much oxygen does
the human lung consume? Anesthesiology 1997; 86: 532–537.
6. Wilkes AR, Mapleson WW. Interference of volatile anaesthetics with infrared analysis of carbon dioxide and nitrous oxide
tested in the Dräger Cicero EM using sevoflurane. British
Journal of Anaesthesia 1996; 76: 737–739.
7. Gravenstein JS, Paulus DA, Hayes TJ, ed. Gas Monitoring in
Clinical Practice, 2nd Edn. Boston: Butterworth-Heinemann,
1995; 113–114.
8. Meriläinen PT. A differential paramagnetic sensor for breathby-breath oximetry. Journal of Clinical Monitoring 1990; 6:
65–73.
9. Theye RA, Michenfelder JD. Whole-body and organ V!O2
changes with enflurane, isoflurane, and halothane. British
Journal of Anaesthesia 1975; 47: 813–817.