Secondary Logo

Journal Logo

Original Papers

Estimation of arterial CO2 partial pressure by measurement of tracheal CO2 during high-frequency jet ventilation in patients with a laryngectomy

Novak-Janković, V.; Paver-Eržen, V.; Fajdiga, I.; Bovill, J. G.; Manohin, A.; Žargi, M.

Author Information
European Journal of Anaesthesiology: January 1998 - Volume 15 - Issue 1 - p 1-5



High-frequency jet ventilation (HFJV), in which inspired gases are injected at high pressure into the airway at rates between 60 and 600 breaths min−1, is the most widely used form of high- frequency ventilation. If the CO2 production and anatomical dead space are assumed to be constant, then as the frequency of ventilation increases, the minute volume required to produce a normal alveolar CO2 concentration increases, while the tidal volume decreases, approaching the volume of the dead space[1]. Adequate alveolar ventilation during HFJV occurs because of the convective transport of gases and increased molecular diffusion [2]. One of the difficulties with HFJV is monitoring the adequacy of ventilation. The arterial CO2 (PaCO2) is the time-honoured and physiologically sound method of assessing the adequacy of ventilation. With controlled ventilation at conventional rates and with normal tidal volumes, PaCO2 can be reliably estimated from end-tidal CO2 (PetCO2). Because of the small tidal volumes delivered during HFJV and the relatively slow response of most CO2 analysers, PetCO2 does not accurately reflect PaCO2[3,4].

One approach to assessing the adequacy of ventilation has been to measure PetCO2 after one or more large breaths following interruption of HFJV [3,5-8]. This is based on the assumption of rapid equilibration between alveolar and pulmonary capillary CO2 partial pressure, so that the PetCO2 from a single large breath following interruption of HFJV should correlate well with PaCO2[8]. This has been demonstrated in dogs [3,5-7] and in patients with impaired pulmonary function [8].

Bourgain and colleagues [9] developed a HFJV which allowed the aspiration of tracheal gas for CO2 analysis (PtCO2) through the injector after stopping the ventilator, but without applying a separate large breath. They used this device in patients undergoing direct laryngoscopy, with the injection catheter inserted through the cricothyroid membrane directly after stopping HFJV. They found a significant correlation between PtCO2 and PaCO2, but PtCO2 underestimated PaCO2 by 0.84 (±0.72 SD) kPa.

A disadvantage of the methods described above is that they either require intubation of the trachea or the use of specially adapted ventilators. We describe our experience with an alternative method for monitoring tracheal CO2 in laryngectomized patients for surgical procedures in the tracheostomy region involving the insertion of a voice prosthesis. For these procedures, HFJV is virtually the only mode of ventilation that does not interfere with the surgeon's work. Intubation of the trachea would not permit the monitoring techniques used in the above-mentioned studies.

Subjects and methods

After obtaining approval from the University of Ljubljana Ethics Committee, Ljubljana, Slovenija, we studied 27 patients with a previous laryngectomy scheduled for insertion of a voice prosthesis. Informed consent was obtained from each patient. The mean (SD) age of the patients was 58 (8.4) years (range 42-71 years) and mean weight 67 (13.5) kg (range 45-98 kg). All the patients had been smokers with clinical evidence of chronic obstructive pulmonary disease. No premedication was given. In the operating theatre, a venous infusion was established and a radial artery was cannulated.

A blood sample for arterial blood gas analysis was collected. At the same time, a catheter (internal diameter 1 mm, length 30 cm) was introduced into the trachea and positioned at a point 1cm above the carina under bronchoscopic control. The distance from the tracheal opening to the tip of the catheter was noted. The catheter was connected to a Dräger (Lübeck, Germany) capnometer with a flow rate of 300 mL min−1. PtCO2 was recorded from the digital read-out of the capnometer. At this flow rate, the monitor response time (10-90%) is 120 ms. The capnometer was calibrated before each patient by aspirating 5% CO2 from a calibration gas bottle. The PtCO2 was measured during spontaneous breathing. After these measurements, anaesthesia was induced with midazolam 2-4 mg i.v., alfentanil 20-40 μg kg−1 and propofol 2 mg kg−1. Vecuronium 0.08-0.1 mg kg−1 was given for neuromuscular block. Anaesthesia was maintained with a continuous infusion of propofol 0.1-0.2 mg kg−1 min−1. Additional vecuronium was given when required.

A plastic 14-gauge, 10-cm long catheter was introduced through the tracheostomy, its tip being positioned 1 cm above the carina under bronchoscopic control. The proximal end of the catheter was fixed to the skin of the chest and connected to the jet injector. High-frequency jet ventilation was started with a Bear jet ventilator (Bear Medical Systems AG, Baar, Switzerland), using a driving pressure of 137.8-275.6 kPa [mean 203.85 (47.57) kPa, an I:E ratio of 0.3-0.4 and a ventilation rate of 100 breaths min−1]. After 15 and 25 min of HFJV, a further blood sample for PaCO2 determination was withdrawn, HFJV was interrupted and the gas sampling catheter was inserted manually into the trachea to 1 cm above the carina and PtCO2 was measured. For each measurement, HFJV was interrupted for about 10-15 s. On completion of the operation, neuromuscular block was reversed with neostigmine 0.03 mg kg−1 and atropine 0.01 mg kg−1. The electrocardiogram, automatic noninvasive arterial blood pressure and oxygen saturation by pulse oximetry were monitored during the procedure.

The PtCO2 measurements were expressed at body temperature and pressure saturated (BTPS). The PaCO2 was measured at 37°C and corrected for the patient's body temperature.

The agreement between the methods was assessed using the method described by Bland and Altman [10]. The bias was calculated as the mean difference between PtCO2 and PaCO2. The upper and lower limits of agreement (precision) were calculated as bias ±2 SD. Comparisons between data obtained during spontaneous breathing and HFJV were made using an Anova and the Student-Newman-Keuls test where appropriate. Statistical analysis was with the Spss statistical package for Windows 6.0; P<0.05 was considered significant.


In four patients, arterial blood gas could not be analysed in the period before anaesthesia. Thus, the data for this period represent only 23 patients. The values of PaCO2 and PtCO2, and the PtCO2-PaCO2 gradient during spontaneous breathing and during HFJV are shown in Table 1.

Table 1
Table 1:
Carbon dioxide partial pressures (kPa) measured in the trachea (PtCO2) and arterial blood (PaCO2), and the PtCO2-PaCO2 differences during spontaneous breathing, and 15 and 25 min after starting high-frequency jet ventilation (HFJV). Values are means (95% CI)

The mean duration of HFJV was 43 min (range 25-70 min). The PtCO2-PaCO2 gradient during spontaneous breathing was significantly lower (P<0.0002) than at either 15 or 25 min during HFJV. Bland-Altman plots for the measurements during spontaneous ventilation and during HFJV are shown in Figs 1 and 2.

Fig. 1
Fig. 1:
Bias analysis of CO2 partial pressures measured in the trachea (PtCO2) and arterial blood (PaCO2) during spontaneous breathing. The bias is calculated as the mean difference between the methods; the limits of agreement are indicated by the mean difference ±2 SD.
Fig. 2
Fig. 2:
Bias analysis of CO2 partial pressures measured in the trachea (PtCO2) and arterial blood (PaCO2) during high-frequency jet ventilation. The points represent data from the combined measurements at 15 and 25 min after the start of ventilation.

During spontaneous breathing the bias was −0.77 kPa (95% CI = −1.00 to −0.54 kPa), and the upper and lower limits of agreement were 0.29 kPa (95% CI = −0.11-0.7 kPa) and −1.83 kPa (95% CI = −2.24 to −1.43 kPa), respectively. During HFJV, the bias was −1.61 kPa (95% CI = −1.76 to −1.46 kPa), and the limits of agreement were −0.48 kPa (95% CI = −0.75 to −0.21 kPa) and −2.74 kPa (95% CI −3.01 to −2.47 kPa).


It is possible that the presence of a laryngectomy in our patients would have altered the pattern of airflow in the lower trachea, sufficient to allow the entrainment of atmospheric air by the capnometer. This would have distorted the relation between CO2 partial pressure in the trachea and arterial blood. In addition, all of our patients had clinical evidence of chronic obstructive pulmonary disease. This could have further contributed to an abnormal PtCO2-PaCO2 gradient because of ventilation-perfusion mismatching and an increased physiological dead space. Therefore, we assessed the agreement between tracheal CO2 and PaCO2 when the patients were awake and breathing spontaneously. During spontaneous breathing, we found that the CO2 partial pressure measured close to the carina underestimated PaCO2 by 0.77 kPa (95% CI = −1.00 to −0.54 kPa). This is comparable with the PaCO2-PetCO2 gradient of 0.65 kPa (4.5 mmHg) reported by Nunn and Hill [11] in healthy, spontaneously breathing, anaesthetized patients. However, our results differ considerably from those of Waldau and colleagues [12], who found a gradient of only 0.10 (0.41) kPa in spontaneously breathing volunteers with PetCO2 measured from a catheter positioned in the hypopharynx. Even smaller gradients have been reported in spontaneously breathing anaesthetized dogs [5].

During HFJV in our patients, the PtCO2-PaCO2 gradient was significantly greater than during spontaneous breathing (P<0.0002). We measured tracheal CO2 partial pressure during interruption of ventilation for 10-15 s, but without giving a large breath. A similar approach was used by Bourgain and colleagues [9], who measured PtCO2 using a modified jet ventilator which aspirated gas samples from the trachea after stopping HFJV. They found a mean(SD) gradient of 1.6 (1.3) kPa in five patients with COPD, whereas the gradient was 0.8 (0.6) kPa in patients without airway disease. Their results in patients with COPD are very similar to our findings. A possible explanation for the increased PtCO2-PaCO2 gradient during HFJV compared with the situation during spontaneous breathing is air entrainment. Our capnometer used a sampling flow rate of 300 mL min−1 and would have aspirated 45-50 mL during the measurement period of 10-15 s. In a patient with a tracheostomy, the anatomical dead space is reduced approximately by half [13]. On average, the anatomical dead space in anaesthetized, intubated patients was 63 mL during artificial ventilation, but only 35 mL during spontaneous breathing [3]. Thus, it is possible that a proportion of the gas aspirated by the capnometer contained entrained atmospheric air, which would dilute the CO2 present. While air entrainment is also possible during spontaneous breathing, it is perhaps less likely. For the measurement of PtCO2 during HFJV, ventilation was stopped and there would be little gas flow along the trachea from the lungs. However, during spontaneous breathing, greater flows would be likely because of active expiration. This would reduce the likelihood of entrainment of external air during the sampling period. During spontaneous breathing, active expiration increases the proportion of alveolar air in the tracheal sample and the PtCO2-PaCO2 gradient is smaller than during interruption of HFJV.

Because of the magnitude of the PtCO2-PaCO2 gradient in our study, and in particular, the amount of variability, we conclude that measurement of PtCO2 using the technique described here gives an unreliable estimate of PaCO2 during HFJV. However, the gradient remained relatively stable during HFJV, with similar values at 15 and 25 min. While the accuracy would have been improved by using a single or multiple breath technique, as described by others, this was not possible in our patients.

Nonetheless, despite its inaccuracy, we believe that the agreement between PtCO2 and PaCO2 is sufficient to allow for adjustment of ventilator settings during jet ventilation, especially when the only alternative may be frequent arterial blood sampling for blood-gas analysis. Obviously, in patients with severe pulmonary diseases, the use of blood-gas analysis is the preferred method for assessing the adequacy of ventilation, but there may be some considerable delay before obtaining the results from the laboratory. The use of our method requires manipulation close to the surgical fields. Where this is not possible or desirable, an alternative to arterial sampling could be transcutaneous PCO2 measurement, but this has a slow response time and the technology is not always available.


1 Sykes MK. Gas exchange during high frequency ventilation. Acta Anaesthesiol Scan 1989; 33 (Suppl. 90): 32-33.
2 Scherer PW, Muller WJ, Raub JB, Haselton FR. Convective mixing mechanisms in high frequency intermittent jet ventilation. Acta Anaesthesiol Scand 1989; 33 (Suppl. 90): 58-64.
3 Sehati S, Young JD, Sykes MK, McLeod CN. Monitoring of end-tidal carbon dioxide partial pressure during high frequency jet ventilation. Br J Anaesth 1989; 63: 47S-52S.
4 Capan LM, Ramanathan S, Sinha K, Turndorf HI. Arterial to end-tidal CO2 gradients during spontaneous breathing, intermittent positive-pressure ventilation and jet ventilation. Crit Care Med 1985; 13: 810-813.
5 Mortimer AJ, Cannon DP, Sykes MK. Estimation of arterial PCO2 during high frequency jet ventilation. Studies in the dog. Br J Anaesth 1987; 59: 240-246.
6Algora-Weber A, Rubio JJ, Dominguez de Villota E, Cortes JL, Gomez D, Mosquera JM. Simple and accurate monitoring of end-tidal carbon dioxide tension during high-frequency jet ventilation. Crit Care Med 1986; 14: 895-897.
7 Mihm FG, Feeley TW, Rodarte AI. Monitoring end-tidal carbon dioxide tensions with high frequency jet ventilation in dogs with normal lungs. Crit Care Med 1984; 12: 180-182.
8 Mason CJ. Single breath end-tidal PCO2 measurement during high frequency jet ventilation in critical care patients. Anaesthesia 1986; 41: 1251-1254.
9 Bourgain JL, McGee K, Cosset MF, Bromley L, Meistelman C. Carbon dioxide monitoring during high frequency jet ventilation for direct laryngoscopy. Br J Anaesth 1990; 64: 327-330.
10Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurements. Lancet 1986; 1: 307-310.
11 Nunn JF, Hill DW. Respiratory dead space and arterial to end-tidal CO2 tension differences in anesthetized man. J Appl Physiol 1960;155: 383-389.
12 Waldau T, Oberg B, Larsen VH. Reliability of CO2 measurements from the airway by a pharyngeal catheter in unintubated, spontaneously breathing subjects. Acta Anaesthesiol Scand 1995; 39: 637-642.
13 Nunn JF. Nunn's Applied Respiratory Physiology, 4th Edn. Oxford: Butterworth Heinemann, 1993: 173.

VENTILATION, high frequency; MONITORING, carbon dioxide, arterial, tracheal, partial pressure

© 1998 European Academy of Anaesthesiology