BJA Advance Access originally published online on December 23, 2005
British Journal of Anaesthesia 2006 96(2):186-194; doi:10.1093/bja/aei302
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
CLINICAL PRACTICE |
Bedside red cell volumetry by low-dose carboxyhaemoglobin dilution using expiratory gas analysis
Department of Emergency Medicine and Critical Care, Saitama Medical Center, 1981 Tsujido-machi, Kamoda, Kawagoe-shi, Saitama 350-8550, Japan
* Corresponding author. Email: sawanom-tky{at}umin.ac.jp
Accepted for publication October 10, 2005.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
Background. We developed a non-invasive, continuous, high-resolution method of measuring carboxyhaemoglobin fraction (COHb%) using expiratory gas analysis (EGA). We assessed whether application of EGA to carboxyhaemoglobin dilution provides red cell volume (RCV) measurement with accuracy equivalent to that of CO-haemoximetry, with a smaller infusion volume of carbon-monoxide-saturated autologous blood (COB).
Method. We assessed the agreement between repeated COHb% measurements by EGA and simultaneous measurement by CO-haemoximetry, using Bland and Altman plot, in healthy subjects and patients with artificially controlled ventilation and no radiological evidence of pulmonary oedema or atelectasis. We assessed the agreement between RCV measurements by EGA with infusion of 20 ml of COB (RCVEGA) and RCV measurements by CO-haemoximetry with infusion of 100 ml of COB (RCVHEM), in healthy subjects.
Results. The ‘limits of agreement’ between COHb% measurement by EGA (1 min average) and CO-haemoximetry were –0.09 and 0.08% in healthy subjects, and –0.11 and 0.09% in patients. Given the resolution of CO-haemoximetry (0.1%), the accuracy of EGA was equivalent to or greater than that of CO-haemoximetry. The ‘limits of agreement’ between RCVEGA and RCVHEM were –0.14 and 0.15 litre. Given the average resolution of RCVHEM (0.14 litre), the accuracy of RCVEGA was equivalent to that of RCVHEM.
Conclusion. EGA provided non-invasive, accurate, continuous, high-resolution COHb% measurements. Applying EGA to carboxyhaemoglobin dilution, we achieved RCV measurements with accuracy equivalent to that of CO-haemoximetry, with one-fifth of the COB infusion volume. However, clinical application of the method is limited to patients with no radiological evidence of pulmonary oedema or atelectasis.
Keywords: blood, volume; measurement techniques, gas exchange; monitoring, carboxyhaemoglobin
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
An accurate, safe, and repeatable method of measuring red cell volume (RCV) at the bedside would provide anaesthesiologists and intensivists with a useful monitoring method for management of patients with anaemia, blood loss, or circulatory failure.
Radioisotope-labelled red blood cell (51Cr-RBC or 99mTc-RBC) dilution is generally accepted as the gold standard for RCV measurement.1 2 However, this method has not been commonly used in clinical applications, due to restrictions in use of radioactive isotopes at the bedside. Carboxyhaemoglobin dilution was introduced as a replacement for 51Cr-RBC. The accuracy of carboxyhaemoglobin dilution has been shown to be equivalent to that of the 51Cr-RBC method,3–5 but requires a bolus infusion of 100 ml of carbon monoxide (CO)-saturated blood, which is potentially hazardous for anaemic or hypovolaemic patients. Ensuring the safety of this method requires reduction of the infusion volume, which necessitates development of techniques with higher resolution in carboxyhaemoglobin fraction (COHb%) measurement.
In the present study, we have developed a novel technique of COHb% measurement by expiratory gas analysis (EGA method), verified its accuracy by assessing agreement with CO-haemoximetry, and assessed whether EGA can provide RCV measurement by carboxyhaemoglobin dilution with accuracy equivalent to that of CO-haemoximetry with a smaller infusion volume.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
Subjects and patients
All protocols were approved by the Ethics Committee of Saitama Medical School.
Subjects for experiments 1, 2, and 4 were 32 healthy adult volunteers (26 male and 6 female; age range, 24–42 yr) who gave written informed consent. No subject participated in more than 1 experiment. Smokers were asked to refrain from smoking for 24 h prior to the experiment.
The subjects for experiment 3 were 12 patients (6 males and 6 females; age range, 21–60 yr) who were admitted to the intensive care unit (ICU) in the emergency department of Saitama Medical Centre. The patient or legal guardian gave written informed consent for the secondary utilization of the clinical data. The reasons for admission were intoxication by narcotics (four patients), cerebral vascular diseases (five patients) and cervical spinal injury (three patients). Ventilation was artificially controlled by endotracheal intubation and a respirator (Drager, EVITA4) for more than 24 h. Their chest radiographs showed no evidence of pulmonary oedema (congestion) or atelectasis.
Experimental settings
The experiments were performed inside an operating room or an ICU equipped with a forced air ventilation system. A plastic cannula was placed in an ante-brachial peripheral vein for sampling and infusion of venous blood. Arterial blood was sampled via a single puncture of a femoral artery or a cannula placed in a radial artery. Arterial oxygen (O2) and carbon dioxide (CO2) tensions (
, 
) and O2 saturation (
) were measured using a blood gas analyser (Radiometer, ABL720), with resolutions of 0.013 kPa (0.1 torr) and 0.1%, respectively. COHb% and other haemoglobin fractions of blood samples were measured using a CO-haemoximeter integrated into the ABL720, with a resolution of 0.1%. Intrapulmonary shunt ratio was estimated using a concentration-based index (FShunt) from these measurements and inspiratory gas O2 concentration (
).6 7 was 0.21 in experiment 2, and ranged from 0.30–0.40 in experiment 3. was measured using an oximeter integrated into the EVITA4, with a resolution of 0.01. We assumed that the difference in O2 concentration between arterial and mixed venous blood was 5.1 ml dl–1.8
The sum of arterial COHb% and oxyhaemoglobin fraction (O2Hb%) was measured using a pulse-oximeter (Nihon Koden, PULSOX-M, Tokyo, Japan). The pulse-oximeter measures the sum as O2 saturation (
), with no discrimination between arterial COHb% and O2Hb%, and with a resolution of 1%.
Expiratory gas and inspiratory gas CO and CO2 concentrations were measured using a Carbolizer (Taiyo, mba2000, Osaka, Japan). The Carbolizer can measure CO and CO2 concentrations continuously (every second), with resolutions of 0.1 ppm and 0.1%, respectively. Two-point calibration of the Carbolizer was performed using a standard gas prior to each experiment. In experiments 1, 2 and 4, subjects were instructed to breathe freely and spontaneously into a device consisting of two one-way valves and sampling tubes. Inspiratory gas and expiratory gas were sampled using this device, for measurement by the Carbolizer. In experiment 3, the sampling device was placed between the endotracheal tube and respirator circuits.
CO-saturated autologous blood (COB) was prepared using the following procedure. The required volume of venous blood was drawn and sealed in a sterile plastic transfusion bag with 100 U of heparin and 400 ml of filtered 100% CO gas. Then, the bag was shaken slowly in an incubator at 37°C for 20 min. Blood samples were drawn from the bag before and after shaking, to confirm that COHb% was greater than 99% and that serum potassium concentration had not increased. This was to ensure that the blood had been saturated with CO without cellular destruction.
Experiment 1
The objective of experiment 1 was to clarify the correlation between expiratory gas CO and CO2 concentrations at a random respiratory phase, with COHb% at a constant value. This analysis was required to develop an algorithm to estimate theoretical end tidal CO concentration (E'CO) from expiratory gas CO and CO2 concentrations. Eight subjects participated in this experiment. Inspiratory gas CO and CO2 concentrations (inCO, inCO2) were determined by averaging continuous measurements from inspiratory gas (room air) obtained over a 60 s period. Then, expiratory gas CO and CO2 concentrations (exCO, exCO2) were measured continuously for 
300 s.
Experiment 2
Based on the results of experiment 1, we developed an algorithm for calculating continuous COHb% in EGA method. The objective of experiment 2 was to assess the accuracy of the EGA by comparing EGA measurements with simultaneous measurements obtained by CO-haemoximetry. Six subjects participated in this experiment. exCO and exCO2 were continuously measured for 1 min to determine baseline. Next, 50 ml of CO-saturated blood was infused. The time at which the infusion was complete (Tinf) was recorded. After the infusion, exCO and exCO2 were measured for 30 min to calculate continuous COHb% by EGA, using the procedure described in the supplementary data to the online version of this article. Venous blood was sampled every 5 min to measure COHb% by CO-haemoximetry. The 1 min averages of COHb% measurements obtained by EGA were calculated for comparison with simultaneous measurements obtained by CO-haemoximetry.
Experiment 3
The objective of experiment 3 was to assess the accuracy of the EGA method for patients with artificially controlled ventilation. Eight patients participated in this experiment. The 1 min averages of COHb% measurements using EGA and simultaneous measurements by CO-haemoximetry were obtained twice with an interval of 12–24 h.
Experiment 4
The objective of experiment 4 was to assess whether COHb% measurement by EGA can provide RCV measurement with accuracy equivalent to that of CO-haemoximetry, and with a smaller infusion volume of CO-saturated blood than CO-haemoximetry. Eighteen subjects participated in this experiment. First, RCV was measured using the EGA method, with infusion of 20 ml of CO-saturated blood (RCVEGA). Baseline COHb% (COHb%bas) was determined by averaging COHb% measurements obtained by EGA over a period of 5 min. Next, 20 ml (Vinf) of COB was infused, and the time of completion of the infusion (Tinf) was recorded. The COHb% (COHb%inf) and haematocrit percentage (Ht%inf) of the CO-saturated blood were measured using the ABL720. The COHb% diminution curve was calculated by exponential regression of continuous COHb% measurements obtained by EGA beginning at Tinf+600 s. Augmented COHb% at Tinf (COHb%aug) was calculated by extrapolation of the COHb% diminution curve to Tinf. RCVEGA was calculated using the following equation:
 |
Statistical analysis
Agreement between COHb% measurements obtained by EGA and CO-haemoximetry in experiments 2 and 3, and between RCVEGA and RCVHEM in experiment 4, was assessed using the method of Bland and Altman.9 In experiments 2 and 3, standard deviations (SDs) of the differences were calculated as repeated measures. For assessment of agreement between RCVEGA and RCVHEM, the resolution of RCVHEM was calculated from the resolution of COHb% measurements obtained by CO-haemoximetry (0.1%), using the following equation:
 |
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
Results of experiment 1
Figure 1 shows the correlation of differences between inspiratory and expiratory gas CO concentration (exCO–inCO) and CO2 concentrations (exCO2–inCO2), at a random respiratory phase. The distribution of y-intercepts of regression lines among the 8 subjects was 0.021 (0.028) ppm [mean (SD)]. Given the resolution of CO measurement (0.1 ppm), the y-intercepts can be considered equivalent to zero. Distribution of the correlation coefficient was 0.984 (0.004) [mean (SD)], indicating strong linear correlation. These results indicate that the ratio of (exCO–inCO) to (exCO2–inCO2) remained constant with fixed COHb%, regardless of the respiratory phase.
|
Results of experiment 2
Figure 2 shows time series data of COHb% measurement by EGA in six healthy subjects, after infusion of 50 ml of COB. Table 1 shows the 1 min average of COHb% measurements by EGA and simultaneous measurements by CO-haemoximetry. In Figure 3, the differences between the two methods of measurement are plotted against their means. The distribution of the differences in 5 or 6 repeated measurements from 6 subjects was –0.01 (0.04%) [mean (SD)]. The ‘limits of agreement’ [mean (2 SD) of the differences] were –0.09 and 0.08%. The distribution of FShunt was 1.5 (0.4%) [mean (SD)].
|
|
|
Results of experiment 3
Table 2 shows the 1 min average of COHb% measurements by EGA and simultaneous measurements by CO-haemoximetry in the patients with artificially controlled ventilation. In Figure 4, the differences are plotted against their means. The distribution of the differences in two repeated measurements from 12 patients was –0.01 (0.05%) [mean (SD)]. The ‘limits of agreement’ were –0.11 and 0.09%. The distribution of FShunt was 5.3 (1.1%) [mean (SD)] and the maximum was 7.3%.
|
|
Results of experiment 4
Table 3 shows the RCV measurements obtained by EGA with 20 ml infusion of CO-saturated blood (RCVEGA), and those obtained by CO-haemoximetry with 100 ml infusion (RCVHEM). In Figure 5, the differences are plotted against their means. The distribution of the differences among the 18 subjects was –0.01 (0.07 litre) [mean(SD)]. The ‘limits of agreement’ were –0.15 litre and 0.14 litre. The distribution of the resolution of the RCVHEM measurements was 0.14 (0.08 litre) [mean (SD)].
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
COHb% measurement by EGA method
There have been several reports detailing estimation of COHb% from expiratory gas CO concentration,10–12 but the resolution of COHb% measurement in those methods was not high enough to replace CO-haemoximetry in RCV measurements using carboxyhaemoglobin dilution. Moreover, a special breathing technique was required in those methods. Thus, continuous measurement of COHb% was impossible, and application was limited to especially cooperative subjects.
We used a Carbolizer mba2000 gas analyser to measure expiratory gas CO and CO2 concentration continuously with high resolution. Experiment 1 revealed that the difference in CO concentration between inspiratory and expiratory gas was in direct proportion to the difference in CO2 concentration, regardless of the respiratory phase. Based on this finding, we developed an algorithm for calculating theoretical E'CO from expiratory gas CO and CO2 concentration at E'CO a random respiratory phase. Continuous COHb% measurement was obtained by converting E'CO to COHb% using Haldane's equation10 (EGA method). When used with a transcutaneous arterial O2 and CO2 tension monitor and a commonly used pulse-oximeter, EGA provides totally non-invasive and continuous COHb% measurement, even for patients who cannot cooperate.
Accuracy and resolution of COHb% measurement using EGA method
Given the resolution of CO-haemoximetry (0.1%), experiments 2 and 3 indicate that the accuracy of EGA (1 min average) was equivalent to or greater than that of CO-haemoximetry, in healthy subjects and in patients with artificially controlled ventilation and no radiological evidence of pulmonary oedema or atelectasis. The resolution of COHb% measurement using EGA is determined by the resolution of CO measurement (0.1 ppm) and the ratio of expiratory gas CO2 to end tidal CO2 The average CO2 ratio was 0.8, which was shown as the concentration of expiratory gas CO–CO2 plots over a relatively high range (Fig. 1). Using the method described in the online supplementary data, we estimated the resolution of EGA at 0.02%, which is five times greater than that of CO-haemoximetry.
Application of COHb% measurement by EGA to red cell volumetry
Continuous COHb% measurement by EGA following infusion of 50 ml of CO-saturated blood accurately represented the two-phase changes of tracer dilution in a one-compartment model13 (Fig. 2). COHb% showed a transient increase exceeding 3.5% during the early mixing phase in some, and did not return to baseline within 30 min in most subjects. This indicates that the infusion of 100 ml of CO-saturated blood, which is required for carboxyhaemoglobin dilution using CO-haemoximetry, induces an increase in COHb% greater than 7% and a further increase in the time required to return to baseline. The reported threshold of COHb% for clinical manifestation is 10% in healthy adults, but is lower in patients with anaemia, hypovolaemia, or cardiovascular disease.14 15 Therefore, the infusion volume must be significantly reduced to ensure safe, repeatable application of carboxyhaemoglobin dilution, because RCV measurement is of primary importance in management of such patients.
As in all tracer dilution methods, the resolution of COHb% measurement must be improved to achieve equivalently accurate RCV measurement with reduced infusion volume. Given that the present average resolution of RCVHEM measurement was 0.14 litre (Table 3), experiment 4 indicates that the accuracy of RCVEGA was equivalent to that of RCVHEM. Thus, application of high-resolution EGA to carboxyhaemoglobin dilution provided accuracy equivalent to that of RCV measurement by CO-haemoximetry, with one-fifth the infusion volume. The reduced infusion volume also shortened the time required for elevated COHb% to return to baseline, and enabled repeated RCV measurement with a shorter interval.
Disadvantage of COHb% measurement using EGA
As with many expiratory gas measurements, COHb% measurement by EGA is based on the assumption that alveolar gas and arterial blood have equivalent CO and CO2 tensions. The agreement between EGA and CO-haemoximetry in experiments 2 and 3 confirms that this assumption is valid for healthy subjects, and for patients with artificially controlled ventilation and no radiological evidence of pulmonary oedema or atelectasis. However, this assumption is clearly not valid for patients with certain respiratory dysfunction, such as ventilation–perfusion (VQ) mismatch in the broad sense of the term, and diffusion impairment.
EGA corrects differences in CO tensions between expiratory and end tidal gas in calculation of COHb%. Similarly, EGA corrects CO differences between end tidal (alveolar) and arterial blood gas as long as CO–CO2 tension ratios are equivalent. We simulated end tidal and arterial CO–CO2 ratios in two extreme patterns of VQ mismatch using a two-alveoli model (Fig. 6A). In the ‘dead space’ pattern, an alveolus is ventilated but not perfused, and the CO–CO2 ratios are equivalent (Fig. 6B). In the ‘shunt’ pattern, an alveolus is perfused but not ventilated, and the CO–CO2 ratios are equivalent only when perfusion of the alveolus (shunt ratio) is very small (Fig. 6C). In the present study, the estimated intrapulmonary shunt ratio of the patients with artificially controlled ventilation was greater than that of healthy subjects, but was smaller than the reported ratios of ICU patients.16 Experiment 3 indicates that, when the shunt ratio is relatively small (<7.5%), the accuracy of EGA in patients with artificially controlled ventilation is equivalent to that of CO-haemoximetry. In cases of diffusion impairment, alveolar and arterial CO–CO2 ratios are not equivalent, because the capacity for diffusion of CO is significantly less than that of CO2.
|
) from ventilation (VX,VY), perfusion (QX,QY), CO and CO2 tensions in alveoli (PAXCO, PAXCO2, PAYCO, PAYCO2) and capillaries (PCXCO, PCXCO2, PCYCO, PCYCO2). (B,C) Two patterns of VQ mismatch. In the ‘dead space’ pattern (B), alveolus X is ventilated but not perfused (QX=0). CO and CO2 tension in alveoli with no perfusion are assumed to be zero (PAXCO=PAXCO2=0). In the ‘shunt’ pattern (C), alveolus X is perfused but not ventilated (VX=0).EGA is expected to provide accurate RCV measurement with reduced infusion volume in subjects with spontaneous or artificially controlled ventilation. However, these disadvantages limit application of EGA to patients with no radiological evidence of pulmonary oedema or atelectasis; that is, no significant diffusion impairment and small intrapulmonary shunt ratio.
|
Supplementary data |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
Supplementary data can be found in the online version of this article at www.bja.oxfordjournals.org.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionSupplementary dataReferences |
|---|
1 International Committee for Standardization in Haematology. Standard techniques for the measurement of red cell and plasma volume. Br J Haematol 1973; 25: 801–14[Web of Science][Medline]
2 International Committee for Standardization in Haematology. Recommended methods for measurement of red cell and plasma volume. J Nucl Med 1980; 21: 793–800
3 Fukui M, Shigemi K. Determination of single and repeated red cell volumes by the indicator dilution method using carbon monoxide as the indicator. Crit Care Med 1989; 17:1199–202[Medline]
4 Obata H, Goto F, Nara T, et al. High predictive value of red cell volume measurement using carboxy-haemoglobin in a rabbit model of haemorrhage. Br J Anaesth 1998; 81: 940–4
5 Ohki S, Kunimoto F, Isa Y, et al. Accuracy of carboxyhemoglobin dilution method for the measurement of circulating blood volume. Can J Anaesth 2000; 47: 150–4[Medline]
6 Wandrup JH. Quantifying pulmonary oxygen transfer deficits in critically ill patients. Acta Anaesthesiol Scand 1995; 39 Suppl 107: 37–44
7 Siggaard-Andersen O, Wimberley PD, Gothgen IH, Siggaard-Andersen M. A mathematical model of the hemoglobin–oxygen dissociation curve of human blood and of the oxygen pressure as a function of temperature. Clin Chem 1984; 30: 1646–51[Abstract]
8 Siggaard-Andersen O, Siggaard-Andersen M. The oxygen status algorithm: a computer program for calculating and displaying pH and blood gas data. Scand J Clin Lab Invest 1990; 50 Suppl 203: 29–45
9 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–10[CrossRef][Web of Science][Medline]
10 Sjostrand T. A method for the determination of carboxyhemoglobin concentrations by analysis of alveolar air. Acta Physiol Scand 1948; 16: 201–7
11 Stewart RD, Stewart RS, Stamm W, Seelen RP. Rapid estimation of carboxyhemoglobin level in fire fighters. JAMA 1976; 235: 390–2
12 Jarvis MJ, Belcher M, Vesey C, Hutchison DC. Low cost carbon monoxide monitors in smoking assessment. Thorax 1986; 41: 886–7
13 Moore FD, Hartsuck JM, Zollinger RM Jr, Johnson JE. Reference models for clinical studies by isotope dilution. Ann Surg 1968; 168: 671–700[Medline]
14 Winter PM, Miller JN. Carbon monoxide poisoning. JAMA 1976; 236: 1502
15 Ellenhorn MJ, Barceloux DG. Carbon monoxide. In: Ellenhorn MJ, Barceloux DG, eds. Medical Toxicology. New York: Elsevier, 1988; 820–9
16 Kjaergaard S, Rees S, Malczynski J, et al. Non-invasive estimation of shunt and ventilation–perfusion mismatch. Intensive Care Med 2003; 29: 727–34[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Adachi, K. Hirota, T. Hara, Y. Sasaki, and Y. Hara Exhaled Carbon Monoxide Levels Change in Relation to Inspired Oxygen Fraction During General Anesthesia Anesth. Analg., September 1, 2007; 105(3): 696 - 699. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||