Eur Respir J

1993, 6, 231-236

Measurement of carbon monoxide transfer

and lung volume in ventilated subjects

P.D. Macnaughton, C.J. Morgan, D.M. Denison, T.W. Evans

Measurement of carbon monoxide transfer and lung volume in ventilated subjects. Depts of Clinical Physiology. Anaesthe-
P.D. Macnaughton, C.J. Morgan, D.M. Denison, T.W. Evans. sia and Intensive Care, Royal Brompton
ABSTRACT: A simple method for measuring lung volume and carbon mon- National Heart and Lung Hospital,
oxide transfer factor (TLco) by a rebreathing technique was assessed in nine London, UK.
healthy volunteers undergoing intermittent positive pressure ventilation (IPPV). Correspondence: T.W. Evans
Measurements of TLco, alveolar volume (V A) and carbon monoxide transfer Depts of Clinical Physiology, Anaesthe-
coefficient (Kco) made at three inspired oxygen concentrations (21, 35 and 70%) sia and Intensive Care
during IPPV were compared to those obtained during spontaneous breathing. Royal Brompton National Heart and
The effects of 10 cmH 2 0 positive end expiratory pressure (PEEP) were also Lung Hospital
studied. Pulmonary capillary blood volume (Vc) and the diffusing capacity of Sydney Street
the alveolar capillary membrane (Dm) were derived. London SW3 6NP UK
There was a close correlation between measurements of TLco during IPPV
(TLC0 1rrv) and spontaneous breathing (TLc0 8 v) (r=0.92). Ventilated TLco was Keywords: Carbon monoxide
diffusing capacity
64±8% of spontaneously breathing TLco. There was a close agreement between lung volume
ventilated and spontaneously breathing measurements of Kco (r=0.95; mean positive end expiratory pressure
difference 0.14, 95% limits of agreement +0.37 to -0.09 mmol·min· 1·kPa· 1·1" 1). Vc positive pressure ventilation
was 92±23 ml during spontaneous breathing and 72±21 ml during IPPV
(p<0.05). PEEP of 10 cmH 20 significantly increased functional residual capacity Received: June 9 1992
(2.3±0.5 to 3.5±0.6 I) and decreased TLco (5.9±1.0 to 5.3±1.2 mmol·min· 1·kPa·•), Accepted after revision October 3 1992
Kco (1.7±0.2 to 1.1±0.3 mmol·min· 1·kPa· 1·1· 1) and Vc (82±22 to 56±20 ml). Dm
did not change with PEEP. PDM was supported by a Medical
This simple method may be a useful means of assessing gas exchange and Research Council Training Fellowship.
lung volume in ventilated subjects. It showed that PEEP increased lung volume
but reduced TLco and that this reduction appeared to be due to a reduction in
capillary blood volume.
Eur Respir J., 1993, 6, 231-236.

The assessment of respiratory function in patients have been complicated and difficult to use at the
ventilated for respiratory failure is frequently bedside. We therefore adapted a standard laboratory
rudimentary and often limited to arterial blood gas rebreathing method for measuring TLco and accessible
analysis [1]. This deficiency has resulted in little be- lung volume, such that it could be undertaken on
ing known of the natural history of the disordered ventilated subjects at the bedside using equipment
physiology of the adult respiratory distress syndrome readily available in any respiratory function laboratory.
and other serious pulmonary disorders. Furthermore, This paper describes the technique and the comparison
assessing the effects of existing and new therapeutic of measurements made in normal volunteers whilst
interventions in such cases has proved difficult. This spontaneously breathing and during positive pressure
is in marked contrast to the thorough evaluation of ventilation. We also used the method to assess the
pulmonary function possible in ambulant patients with effects of positive pressure ventilation, with and with-
less severe respiratory impairment. Thus, the meas- out 10 cmH 20 positive end expiratory pressure
urement of pulmonary carbon monoxide transfer (PEEP), on lung volume and gas exchange in normal
capacity (TLco); or carbon monoxide diffusing capac- subjects.
ity, (DLco) is commonly used as a simple method of
assessing efficiency of pulmonary gas exchange. TLCO
could be a useful method of assessing patients with Methods
acute respiratory failure requiring positive pressure
ventilation, but few descriptions of methods for Nine healthy male volunteers (age range 25--40 yrs,
measuring TLco in ventilated subjects have been pub- smoker) who gave their informed consent were
lished. Those techniques that have been described recruited. The study was approved by the Ethics
 232 P .D. MACNAUGHTON ET AL.

Committee of the Royal Brompton National Heart and Haemoglobin (Hb) and carboxyhaemoglobin (COHb)
Lung Hospital. levels were measured simultaneously by eo-oximetry
(Coming 2500, Braintree, Essex, UK).
All measurements were undertaken with the subjects
Rebreathing measurement in the supine posture. Two different bag volumes
were used; 80% of supine vital capacity for sponta-
A laboratory method for the measurement of TLco neously breathing measurements (i .e. about 3,500 ml)
by rebreathing was adapted for use in ventilated pa- and I ,000 ml for ventilated measurements. This rather
tients [2]. In brief, a rebreathing bag with an attached small volume was chosen for measurements in venti-
two-way valve was filled from a calibrated syringe lated subjects as it was predicted that this would be
with a measured volume of gas comprising helium the largest volume that would be tolerated by most
(He) 14%, carbon monoxide (CO) 0.3%, oxygen (0 2) patients with lung disease undergoing mechanical
either 21%, 35% or 70%, and balance nitrogen (N 2) . ventilation.
The initial concentrations of these gases were meas-
ured with a combined katharometer and infra-red
analyser (PK Mor·gan, Kingston, Surrey, UK) and an Bag-in-box system
anaesthetic respiratory gas monitor (5250 RGM,
Ohmeda, Hatfield, Herts, UK). At the end of a full A bag-in-box system was designed and built, to
expiration to residual vo lume (RV) when spontane- permit the application of these rebreathing manoeuvres
ously breathi ng, or at the end of a normal expiration to ventilated subjects without influencing set
(functional residual capacity (FRC)) when ventilated, ventilatory pressures, such as the level of PEEP. The
the valve was switched to the rebreathing position. system was placed in the breathing c ircuit between
Rebreathing continued for a timed period of six patient and venti lator (fig. I). A s ingle manually-
breaths (approximately 15 s during spontaneous operated va lve swi tched between conventional venti-
breathing, 25 s if ventilated), ensuring that the lation and rebreathing mode. The box contained a 1
bag emptied completely with each inspiration. l rubber anaesthetic bag with a filling port on the
The duration of rebreathing was taken from the valve block. Each subject was trained to tolerate
moment of valve opening until the valve was closed positive pressure ventilation (Drager Evita, Drager,
at the end of the sixth expiration. The concentrations Hemel Hemstead, Herts, UK) via a mouthpiece,
of He, CO, 0 2 and C0 2 in the breathing bag were then whilst wearing a noseclip. The ventilator was set to
measured. The end expiratory He and CO concentra- deliver a tidal volume of 1,000 ml, at a rate of 15
tions were corrected for changes in 0 2 and C0 2 breaths·min-1, and with an inspiratory to expiratory
concentrations which occurred during the re- time ratio of 1:1.5. The rebreathing bag was filled and
breathing test. To estimate "CO back pressure" (C0 5 p) placed in the respiratory circu it as described above.
a 2 ml venous blood sample was obtained im- After a period for stabilization, the valve was switched
mediately prior to each series of measurements at the end of a tidal expiration (FRC) and rebreathing
and 10 m in after the completion of the last test. continued for six breaths as described above.

Rebreathing
Filling port

Non-rebreathing

Outlet
to Rebreathing bag
patient

Inlet from ventilator

Fig. 1. - Bag-in -box system.
 CARBON MONOXIDE TRANSFER DURING VENTILATION 233

Experimental protocols concentration during each test was calculated using the
following formulae:
After an initial rest period of 10 min, measurements
of TLco were undertaken at intervals of 7.5-10 min, (COHbr.nal-COHb;n;,;.,)
in order to facilitate adequate wash-out of alveolar CO COHb,= COHbinitial +x - -- - --
and He between tests. Each result represented the n
mean of two measurements, which were repeated if
there was a greater than 10% difference between them. Where x = test number and n = total number of tests
Two series of experiments were undertaken. undertaken between measurements of COHb. COBP
was then derived as predicted by the Haldane relation
[5]:
Experiment 1

A series of paired measurements was made to com-

pare results obtained from the standard spontaneously- 225xHb0 2
breathing laboratory rebreathing technique with those
from the bag-in-box system in ventilated subjects. In Where Pco 2 (mean capillary oxygen tension) was
order to study a range of gas transfer values in normal estimated to be equal to the end expiratory value -
volunteers, measurements of accessible lung volume 1.5% and Hb0 2= 1-COHb. TLco was corrected to a
(V A) and TLco were undertaken by seven of the sub- haemoglobin of 14.6 g·dl·' according to the method of
jects, spontaneously breathing and ventilated using gas CoTES et al. [6].
mixtures containing 0 2 concentrations of 21, 35 and From RouoHTON and FoRSTER [3], TLco can be
70%. In the other two subjects, measurements were expressed as the sum of two conductances:
undertaken with 21% 0 2 only. Subjects breathed the
equivalent fractional inspired oxygen concentration
(Fro 2) via a non-rebreathing circuit for 3 min prior to --= - + -
each measurement. Each pair of measurements was TLCO Dm eve
made on separate days and in identical conditions.
By measuring TLco at three different oxygen 8 is derived from the Hb and the alveolar oxygen
concentrations, values could be derived for the two partial pressure, according to the following formulae
components which contribute to the transfer of carbon [3]:
monoxide, diffusion across the alveolar capillary
membrane (Dm), and the chemical reaction with the
blood (8Vc, where 8=reaction constant for combina- = - ---------
tion of carbon monoxide with haemoglobin; and 8 (Hb/14.6)x(1-COHb)
Vc=volume of pulmonary capillary blood) [3]. Values
for Dm and Vc were obtained for seven subjects while Dm and Vc were then obtained from the y axis inter-
breathing spontaneously, and while undergoing cept and the slope, respectively, when 1/TLco was
mechanical ventilation. plotted against 1/8.

Experiment 2 Statistical methods

The effect of PEEP on lung volume and TLco Results are shown in the text and figures as mean
was assessed in eight ventilated normal subjects. ±standard deviation. Correlation between measure-
Measurements were made at 0 and 10 cmH 20 of ments was assessed using linear regression analysis.
PEEP, the level of which was confirmed from the The mean difference between ventilated and non-
displayed airway pressure. For each level of PEEP, ventilated measurements and the 95% limits of agree-
measurements were undertaken at three different levels ment were calculated for Kco and TLco [7) .
of Fro 2 (21, 35 and 70%). The order of measurements Student's paired t-test was used to analyse changes in
was randomized. Lung volume, TLco, Vc and Dm the measured parameters following the application of
were derived for each level of PEEP. PEEP. A value of p<0.05 was considered to be
significant.

Calculations
Results
The alveolar volume (V A), end expiratory volume
(either FRC or RV depending on the rebreathing ma- A total of 23 paired measurements of TLCO and Kco
noeuvre undertaken), TLco and carbon monoxide in nine subjects was made to compare the spontane-
transfer coefficient (Kco) were calculated using ously breathing and ventilated methods. Ventilated
standard equations [4] corrected for C0 8 p. The COHb TLCO (TLC01ppv) was 64±8% of spontaneously breathing
 234 P.D. MACNAUGHTON ET AL.

TLco (TLCOsv), but there was a significant correlation The mean Vc fell from a baseline value of 82±22 to
between the two (r=0.92, p<O.OOI) (fig. 2). There was 56±20 m! following the application of PEEP (p<0.05).
also a significant correlation between ventilated Kco There was a significant correlation between the
(Kco1rpv) and spontaneously breathing Kco (Kco 5 v) change in TLCO and the change in Vc for individuals
(r=0.95, p<0.001) (fig. 3). Kco 1rpv tended to be (r=0.71, p<0.05). PEEP had no effect on Dm;
higher than Kco5 v; the mean difference between the 7.5±1.0 mmol·min- 1·kPa- 1 without PEEP and 7.2±1.5
two being 0.14 (95% limits of agreement +0.37 to mmol·min- 1·kPa- 1 with 10 cmH 20 PEEP (p=0.27).
-0.09) mmol·kPa- 1-min- 1·[- 1•
In seven subjects, Dm and Vc were derived from
measurements of TLCOsv and TLC01ppv made at three Discussion
different values of Fro 2 • Mean V c was 92±23 m!
during spontaneous breathing and 72±21 m! (p<0.05) We have shown that TLco and lung volume can be
during mechanical ventilation. Dm measured during measured in ventilated subjects using a simple
spontaneous breathing and during mechanical ventila- rebreathing technique, and that PEEP results in an
tion was 14.3±1.7 and 8.1±1.1 mmol·min 1·kPa- 1, increase in V A associated with a reduction in TLco.
respectively (p<0.001). Although TLCO can be measured using a number of
The application of 10 cmH2 0 of PEEP caused a rise methods, the single breath technique is the commonest
in mean functional residual capacity (FRC) (2.3±0.5 in routine clinical use. However, the rebreathing
to 3.5±0.6 l, p<0.001). TLco fell from 5.9±1.0 to method would appear to be the most appropriate for
5.3±1.2 mmoJ.kPa- 1·min- 1 (p=O.Ol) and Kco fell from undertaking measurements in ventilated subjects for a
1.7±0.2 to 1.1±0.3 mmol·min- 1-kPa- 1·/ 1 (p<O.OOl). number of reasons. Firstly, the single breath technique
is likely to be influenced to a greater extent by the
9.0 marked inhomogeneity of gas distribution frequently
"';-
Line of identity •• encountered in the critically ill, when it would
c: underestimate the true value of TLco. Secondly, it is
-~ 7.8
"';- much easier to develop a rebreathing method which is
•
...•
et!
Cl.. sterile and reduces the risks of cross infection [2].
~
6.6 Thirdly, a close relationship between rebreathing and
0
E • single breath measurements has been demonstrated [2].

..
E
5.4 • Finally, the single breath method would probably be

• • •••
>
0...
0...
poorly tolerated by critically ill subjects, who may be
0- unwilling or unable to breathhold for 10 s .
0
--'
I- 4.2 In calculating TLco from only two measurements of
•• • CO concentration, we assumed that the lung is a single
compartment and that instant gas mixing occurs
3.0
3.0 6.0 9.0 12.0 15.0 throughout the compartment, such that the elimination
of CO follows a single exponential pattern. These
TLCOsv mmol·kPa·1·min-1 assumptions result in the measured value of TLco
underestimating the true value. It has been proposed
Fig. 2. - TLco spontaneously breathing (TLC05 v) plotted against
TLco ventilated (TLco 1PPv); TLco 1PPv=0.598·TLco 5 v+0.28; r=0.92; that CO uptake during rebreathing should be consid-
p<O.OOI. TLco: transfer factor of the lungs for carbon monoxide. ered as a two-compartment model [8], described
mathematically as a biexponential process; a fast
component representing distribution between the
2.5
"';-
• rebreathing reservoir and alveolar compartment, and a
slow component representing uptake into the pulmo-
~
"';-
et! 2.0 :-•• Line of identity
nary capillary blood. However, in order to measure
Cl..
~
• • • • TLco using this analysis, continuous measurements of
"';-
c:
-~ 1.5 •
•••• • CO concentrations are necessary. Mass spectrometry
and the stable isotope of CO, 12 C 180, are needed as the
•• •
0
E abundant species of CO (1 2C 16 0) and atmospheric
E 1.0 nitrogen have the same mass:charge ratio and are,
>
0...
0...
therefore, indistinguishable [9]. As making these
0-
0 0.5 measurements is not feasible in the majority of inten-
::.::
sive care units, we assessed this simpler method of
measuring TLco using equipment readily available in
0
0 0.5 1.0 1.5 2.0 2.5 most centres, whilst recognizing the limitations and
potential errors of the technique.
Kco SV mmol·min- 1·kPa· 1·t· 1 A second assumption made, is that the system vol-
Fig. 3. - Kco ventilated (Kco,PPvl plotted against Kco spontane-
ume (bag+lung) remains constant during the period of
ously breathing (Kco5 v); Kco 1PPv=0.919·Kco 5 v+0.26; r=0.95; p<O.Ol. rebreathing. The uptake of CO and the highly
Kco: transfer coefficient for carbon monoxide. insoluble He can be ignored due to the very small
 CARBON MONOXIDE TRANSFER DURING VENT! LATION 235

volumes involved. The increase in C0 2 concentration the same technique, and is lower than the result
of about 4% during the period of rebreathing is obtained using a recently described technique for the
matched by a similar fall in 0 2, reflecting a respira- simultaneous measurement of CO and nitric oxide
tory quotient of approximately 1. The system volume, (NO) transfer factors [16). The exact value of Dm
therefore, remains essentially constant using the gas obtained by the technique used in the current study is
mixture employed in this study. However, as the very dependent on the value of 8 (a composite rate
analysers used to measure CO and He have a soda constant for the combination of carbon monoxide with
lime filter to remove any C0 2 , the post-rebreathing haemoglobin within the erythrocyte). A number of
measurements must be corrected for this loss. Finally, different values have been described which may
in order to simplify the calculation of TLco the effec- explain some of the differences seen in estimates of
tive rebreathing rate is assumed to be infinity. In Dm between studies. Furthermore, Dm is calculated
reality, by using a rate of 15 breaths·min- 1 the true from the y intercept of a regression line and is likely
value of TLco tends to be underestimated. However, to have a greater error than the measurement of Ye
the use of higher respiratory rates in ventilated subjects represented by the slope of the same line.
generated unacceptably high airway pressures and may The theory behind measuring V c and Dm using
also have adverse effects on cardiac output. different values of F10 2 assumes that the haemoglobin
TLC01rrv correlated closely with TLCOsv• although the is almost fully saturated along the length of the pul-
former tended to be approximately 60% of the self- monary capillary, such that each molecule has only
ventilating value. This reflects the fact that ventilated one potential binding site for CO (i.e. the haemo-
measurements were made at a mean alveolar volume globin is either Hb0 3 or Hb0 4 ). As the value of 8
of 3.22 l compared with 5.69 l for TLCOsv· It is well- depends upon the number of oxygen molecules bound
known that TLCO tends to fall as VA drops [10, 11]. to haemoglobin [4), under these conditions it can be
This may reflect a true fall in the efficiency of gas assumed to be constant. In theory, measurements
exchange, due to a reduction in Ye and alveolar using this technique should only be undertaken using
surface area, combined with impaired matching of gas mixtures containing more than 28% oxygen [5),
ventilated to perfused alveoli at low lung volumes. In when it can reasonably be assumed that only Hb0 3 or
addition, it may be an a1tifact due to an increased dead Hb0 4 are present. However, a number of previous
space:tidal volume ratio, such that when a small studies have utilized measurements with a Fro 2 of 21%
volume is inhaled a larger proportion of the gas mix- and have still obtained reasonable values for Ye and
ture is in contact with a non-gas exchanging surface. Dm [13-15]. Moreover, we found no significant
There was a close agreement between the ventilated difference between Vc and Dm calculated from meas-
and spontaneously breathing values for Kco, the urements made in this study using 35 and 70% oxygen
former being only 0.14 mmol·min- 1·kPa-1·l- 1 greater than only, or from those values derived when all three
the latter. Measured Kco increases if estimations are oxygen levels were used.
made at a small alveolar volume [ 10, 11] as V c Mean Ye during IPPV was significantly lower than
decreases less than the fall in lung volume. It might during spontaneous ventilation. This may reflect the
have been predicted that Kco 1rrv would be consider- increase in intrathoracic pressure during IPPV, result-
ably larger than Kco 5 v, as the ventilated measurements ing in a reduced capillary volume, or may reflect the
were undertaken at a smaller VA. However, raised lower V A at which the ventilated measurements were
intrathoracic pressure during positive pressure ventila- made. The large reduction in Dm measured during
tion reduces Kco by means of a fall in Ye [12]. IPPV reflects the large differences in V A between
These effects would appear to cancel each other out, ventilated and spontaneously breathing measurements.
such that Kco 1rrv and Kco 5 v agree closely with each By employing a bag-in-box system, we were able to
other. The close correlation between the ventilated undertake measurements without interfering with the
and spontaneously breathing measurements implies that set ventilator pressures and were, therefore, able to
when TLco and Kco are measured by this technique, assess the effects of PEEP. The application of 10
they reflect the same changes in lung function and cmH 20 of PEEP caused a highly significant increase
respond to the same influences as the more conven- in FRC in all subjects. Assuming that the subjects had
tional measurements. a normal total thoracic compliance of around 100
The technique of assessing Dm and Ye by measur- ml·cmH 20- 1, the observed mean increase in FRC of
ing TLco at different values of F10 2 was first 1.1 l was as predicted. PEEP also caused a signifi-
described by ROUGHTON and FORSTER (3). It has been cant fall in TLC0 1rrv in all but one subject studied.
shown to be both accurate and reproducible, and has When combined with the increase in lung volume, this
been used to assess change in Ye in a variety of caused Kco to fall in all subjects. The subject who
pathological states including emphysema [13), rheu- failed to show a fall in TLco with PEEP was the only
matoid arthritis [14), and pulmonary embolism [15). smoker studied, and may have had ventilation:
The mean value of 92 ml obtained for Ye measured perfusion mismatch which was reversed with PEEP.
during spontaneous respiration is similar to values re- PEEP caused a significant fall in Ye but had no
ported in other studies of normal subjects. The mean effect on Dm. As there was a positive correlation
value of 14.3 mmol·min- 1·kPa- 1 for Dm in this study is between the change in TLco and change in Ye, the
at the lower end of the range reported by others using effect of PEEP on TLco was probably due to a fall in
236 P.D. MACNAUGI-ITON ET AL.

Ye. Dm results were much more variable (see above) assessing agreement between two methods of clinical
and no clear trend emerged following the application measurement. Lancet 1986; i: 307-310.
of PEEP. 8. Hook C, Meyer M. - Pulmonary blood flow, diffus-
ing capacity and tissue volume by rebreathing: theory.
Respir Physiol 1982; 48: 255-279.
9. Burchardi H, Stokke T. - Pulmonary diffusing ca-
References pacity for carbon monoxide by rebreathing in mechanically-
ventilated patients. Bull Eur Physiopathol Respir 1985;
I. Macnaughton PD, Morgan CJ , Denison DM, Evans 21(3): 263-273.
TW. - Pulmonary function testing in the intensive care 10. Mcgrath MW, Thompson ML. - The effect of age,
unit. Respir Med 1990; 84: 437-443 . body size and lung volume on alveolar-capillary permeabil-
2. Denison DM, Cramer DS, Hanson P. - Lung function ity and diffusing capacity in man. 1 Physiol 1959; 146:
testing in AIDS. Respir Med 1989; 83: 133-138. 572-582.
3. Roughton FJW, Forster RE. - Relative importance of 11. Denison D, Al-Hillawi H, Turton C. - Lung func-
diffusion and chemical reaction rates in determining rate of tion in interstitial lung disease. Semin Respir Med 1984; 6:
exchange of gases in the human lung, with special reference 40-54.
to true diffusing capacity of pulmonary membrane and 12. Burgess JH, Gillespie J, Graf PD, Nadel JA. - Ef-
volume of blood in the lung capillaries. 1 Appl Physiol fect of pulmonary vascular pressures on single breath carbon
1957; 11(2): 290-302. monoxide diffusing capacity in dogs. 1 Appl Physio/ 1968;
4. Cotes JE. Measurement of the transfer factor 24: 692-696.
(diffusing capacity) for the lungs and its subdivisions. In: 13. Morrison NJ, Abboud RT , Muller NL , et al .
Lung Function: Assessment and Application in Medicine. Pulmonary capillary blood volume in emphysema. Am Rev
4th ed. Oxford, Blackwell Publications, 1979; pp. 230- Respir Dis 1990; 141: 53-61.
250. 14. Hills EA, Geary M. - Membrane diffusing capacity
5. Forster RE. - Diffusion of gases across the alveolar and pulmonary capillary volume in rheumatoid disease .
membrane. In: Fenn WO, Rahn H, eds. Handbook of Thorax 1980; 35: 851-855.
Physiology. Respiration. Bethesda, Maryland, American 15 . Sharma GV, Burleson VA, Sasahara AA. - Effect of
Physiological Society, 1987; pp. 71-88. thrombolytic therapy on pulmonary capillary blood volume
6. Cotes JE, Dabbs JM, Elwood PC, et al. Iron in patients with pulmonary embolism. N Engl 1 Med 1979;
deficiency anaemia: its effect on transfer factor for the lung 303: 842-845.
(diffusing capacity) and ventilation and cardiac frequency 16. Guenard H, Varence N, Vaida P. - Determination of
during submaximal exercise. Clin Sci 1972; 42: 325- lung capacity blood volume and membrane diffusing
335. capacity by measurement of NO and CO transfer. Respir
7. Bland JM, Altman DG. Statistical methods for Physiol 1987; 70: 113-120.