Secondary Logo

Journal Logo


Intermittent or Continuous Carbon Dioxide Insufflation for De-Airing of the Cardiothoracic Wound Cavity? An Experimental Study with a New Gas-Diffuser

Svenarud, Peter MD*; Persson, Mikael MSc*†; van der Linden, Jan MD, PhD*

Author Information
doi: 10.1213/00000539-200302000-00005
  • Free


Insufflation of carbon dioxide gas (CO2) into the wound cavity during open-heart surgery may reduce peri- and postoperative neurological and myocardial complications (1–6) because CO2 dissolves in blood and tissues ≥25 times faster than air (7,8). Furthermore, CO2 is 50% heavier than air, which facilitates air displacement in the wound cavity.

The commonly used insufflation device is a thin open-ended tube, but its ability to provide efficient air displacement has recently been questioned (9). Earlier experimental and clinical studies of CO2 de-airing in cardiac surgery have neither investigated the aspects of intermittent CO2 insufflation nor examined the influence of an enlarged cardiothoracic wound cavity volume due to an open pleural cavity.

The first aim of the study was to investigate the degree of air displacement during steady-state with a new insufflation device, a gas-diffuser, at different CO2 flows in a cardiothoracic wound cavity model. A conventional open-ended tube served as a control. Second, we studied air displacement at the start of and after discontinuation of CO2 insufflation with the gas-diffuser. Third, we studied the influence of an open pleural cavity on the air displacement.


The new gas-diffuser (patented by Cardia Innovation AB, Stockholm, Sweden) consists of a fixable polyvinyl chloride tube with an inner diameter of 2.5 mm and a diffuser (14 × 18 mm) at the end of the tube (Fig. 1). The diffuser has a cylindrical shape and consists of soft polyurethane foam with open cells (density, 30 kg/m3). A gas jet insufflated into the foam will be diverted into all directions already within the diffuser so that the gas exits uniformly from the diffuser surface. The chosen size of the diffuser surface is much larger than the inner cross-sectional area of the 2.5-mm tube opening, and the velocity of the gas jet is accordingly substantially reduced. The function is similar to the distribution of air during inhalation through the trachea and bronchi and down to the alveoli. The high air flow velocity in the relatively small cross-sectional area of the trachea is reduced to laminar flow with a very low flow velocity within the large area of alveolar tissue. An open-ended tube with an inner diameter of 2.5 mm served as a comparative control (Fig. 1).

Figure 1
Figure 1:
The new gas-diffuser consists of a fixable polyvinyl chloride tube with an inner diameter of 2.5 mm and a soft polyurethane diffuser (14 × 18 mm) at the end of the tube. An open-ended tube with an inner diameter of 2.5 mm served as a control.

The CO2 flow was measured with a back-pressure-compensated oxygen (O2) flowmeter, because a flowmeter for medical CO2 was unavailable at the time of the study. The O2 reading scale was adjusted for CO2 by a universal flowmeter (ABB/Fisher & Porter, Göttingen, Germany) because of the higher density of CO2 gas. The universal flowmeter consisted of a measuring tube (FP 1/4-16 G-5/81) with a spherical stainless-steel float (SS-14). The universal flowmeter was not used for measurements in the study because of its lack of back-pressure compensation. This problem was avoided during the calibration by measuring the CO2 outflow distal to the end of the insufflation device. The reading scale of the universal flowmeter was calculated for the used gas (medical CO2; AGA Gas AB, Stockholm, Sweden) at 20°C and at 760 mm Hg with a computer program (FlowSelect Version 2.0; ABB/Fisher & Porter).

Air displacement in the wound cavity model was assessed by analyzing the remaining air content (%Air), which is given byMATH where %O2 is the measured O2 concentration and %O2(ref) is the O2 concentration in atmospheric air near sea level (20.95%) (10). The O2 concentration was measured with an O2 sensor (CheckMate 9900; PBI Dansensor, Ringsted, Denmark), which has a gas sampling volume of <2 mL, a response time of <2 s (>20.95% change in O2 concentration in both directions), a range of measurement of 0.0001% to 100% O2, and an accuracy of 1% of the measured value. The sampling probe was a 1.5-mm-thick Teflon tube. The O2 instrument was connected to a personal computer for the recording of data.

Air displacement was studied in an anatomical torso model with an open cardiothoracic wound containing a silicone replica of the heart and great vessels (Fig. 2). The shape of the model was based on the maximal measurements of the open chest wounds of five adults undergoing cardiac surgery (standard sternotomy and during cardiopulmonary bypass with an empty heart). We presupposed that a wound cavity with a large opening would be more difficult to de-air because of increased diffusion. The torso was placed horizontally on the operating table of a normally ventilated operating room for cardiac surgery (the downward laminar airflow from the ceiling above the operating table was approximately 2500 m3/h). The wound opening was 20 cm long (midline) and 12 cm wide. The volume of the wound cavity without the artificial heart was 2.5 L. The external volume of the artificial heart, including the great vessels, was 1.0 L, giving a residual cavity volume of 1.5 L. Furthermore, the torso cavity could be extended with an additional volume of 2.5 L, corresponding to an opened left pleural cavity with a collapsed lung. The orifices of the insufflation devices were positioned 5 cm below the wound opening adjacent to the diaphragm. The tube and the gas-diffuser were pointed toward the center of the wound cavity and not toward the site of O2 mea-surements. CO2 was insufflated at a flow of 2.5, 5, 7.5, and 10 L/min. The remaining air content was measured at the topmost part of the right atrium, 5 cm below the wound opening, close to the site of the atrial incision in mitral valve surgery.

Figure 2
Figure 2:
Air displacement was studied in an anatomical torso model with an open cardiothoracic wound containing a silicone replica of the heart and great vessels. The shape of the model was based on the maximal measurements of the open chest wounds of five adults undergoing cardiac surgery.

The air displacement efficiency of the two insufflation devices was assessed during steady state. A stable O2 concentration was considered to be present when values fluctuated around a constant value over 30 s. Thereafter, the O2 concentration was recorded 10 times in succession, once every 5 s. Furthermore, the air content was recorded every 5 s during the first 60 s of initial CO2 filling and after termination of continuous CO2 insufflation, by using the gas-diffuser. These recordings were repeated 10 times. All measurements were conducted with and without an open left pleural cavity. The remaining CO2 in the model was removed with the help of a rough sucker before every change of CO2 flow or insufflation device, whereupon air movements around the model were left to settle for 1 min.

Differences were considered statistically significant if P < 0.05. Data are presented as medians and ranges. Mann-Whitney U- and Wilcoxon’s tests were used whenever appropriate. Pairwise comparison analysis of variance (ANOVA) was used for comparison between groups with repeated measures. ANOVA with Bonferroni’s correction was used for multiple comparisons within a group.


Figure 3 depicts the air content (steady state) in the model with a closed (Fig. 3A) and open (Fig. 3B) left pleura when CO2 was insufflated through the 2.5-mm tube and the gas-diffuser. When the cavity was insufflated with CO2 through the 2.5-mm tube, the median air content was between 82.0% and 88.2% (range, 78.8%–91.2%) at the studied CO2 flows, including both a closed and an open left pleural cavity. With the gas-diffuser, the air content was statistically lower (P < 0.001) than with the 2.5-mm tube at all studied CO2 flows. This significant difference between the two insufflation devices appeared when the left pleura was closed as well as when it had been opened. The median air content was 6.9% (range, 6.6%–7.3%) at a CO2 flow of 2.5 L/min when the gas-diffuser was used and when the pleural cavity was closed. When the CO2 flow was increased to 5 L/min under similar circumstances, the corresponding figure was 0.65% (range, 0.54%–1.3%;P < 0.001). A further decrease (P < 0.001) in median air content to 0.38% (range, 0.37%–0.42%) was seen when the CO2 flow was increased to 7.5 L/min. At a CO2 flow of 10 L/min, the median air content was 0.29% (range, 0.27%–0.33%;P < 0.001). With an open left pleural cavity, the corresponding median air contents were 7.2% (range, 6.8%–8.0%), 0.47% (range, 0.38%–0.53%), 0.37% (range, 0.34%–0.38%), and 0.18% (range, 0.16%–0.20%), with statistically significant differences between all flows (P < 0.001). At a CO2 flow of 2.5 L/min, the air content was somewhat higher with an open than with a closed pleura (P < 0.001), whereas with higher CO2 flows, the air content was slightly higher with a closed pleura (P < 0.003).

Figure 3
Figure 3:
Median and range of air content in a cardiothoracic wound cavity model with a closed (A) and open (B) pleural cavity insufflated with a 2.5-mm tube and a gas-diffuser at 2.5, 5, 7.5, and 10 L/min of CO2.

Figure 4 depicts the air content in the model with a closed (Fig. 4A) and an open (Fig. 4B) left pleural cavity during the first 60 s of initial CO2 filling, when CO2 was insufflated with the gas-diffuser at 2.5, 5, 7.5, and 10 L/min. Paired comparison of air contents with repeated-measures ANOVA showed statistically significant differences (P < 0.001) among the four CO2 flows, both with an open and a closed left pleural cavity. Statistically significant differences (P < 0.001) also appeared in comparisons between the open and closed left pleural cavity at the same flows. With the pleura intact, a CO2 flow of 10 L/min resulted in the quickest decrease in air content, and a stable low value was reached 20 s after the start of CO2 filling (tested with ANOVA, including Bonferroni’s correction). At this point in time (20 s after the start), the air content showed statistically significant differences among the 4 CO2 flows (P < 0.001). The air content reached a stable low level after 25, 50, and 55 s at a CO2 flow of 7.5, 5, and 2.5 L/min, respectively. The corresponding stable air content with an open pleural cavity was achieved after 30 and 35 s at a CO2 flow of 10 and 7.5 L/min, respectively. Stable low values were not obtained with a CO2 flow of 5 or 2.5 L/min (open pleura) within 1 min of CO2 filling. At each of the CO2 flows used, the air content was statistically lower after 20 s of CO2 filling, when the left pleural cavity was closed, than when it was open (P = 0.001).

Figure 4
Figure 4:
Median and range of air content in a cardiothoracic wound cavity model with a closed (A) and open (B) left pleural cavity insufflated with the gas-diffuser at 2.5, 5, 7.5, and 10 L/min of CO2 during the first 60 s of CO2 filling.

Figure 5 shows the air content in the model, with and without an open left pleural cavity filled with CO2, during the first 60 s after the CO2 supply was discontinued. In paired comparisons with repeated-measures ANOVA, the air contents did not differ statistically (P = 0.49) with a closed or open left pleural cavity. However, the air content increased (P < 0.01) between every 5-s interval, both with an open and a closed left pleural cavity (tested with ANOVA, including Bonferroni’s correction), except between 0 and 5 s with a closed pleural cavity and between 0 and 10 s with an open pleural cavity.

Figure 5
Figure 5:
Median and range of air content in a cardiothoracic wound cavity model with a closed and open left pleural cavity during the first 60 s after termination of CO2 insufflation.


Most surgeons try to prevent air embolism during open-heart surgery. Usual surgical measures include atrial venting, vent suction, Trendelenburg position, ventricle emptying by compression, evacuation of trapped air (diagnosed by transesophageal echocardiography) by gravitation or aspiration, and insufflation of the cardiothoracic wound with CO2. Earlier experimental animal studies (1–6) have shown that arterial embolization of air (injected) into the brain and the heart may not only cause cerebral and myocardial dysfunction, but may also lead to convulsions, infarctions, ventricular fibrillation, and death. In contrast, injection of CO2 is much better tolerated (1–6). We are aware of only a single study in which insufflation of the cardiothoracic wound with CO2 was advantageous. In an experimental study with dogs, Eguchi et al. (2) exposed the mitral valve to the open cardiothoracic wound by opening the left atrium widely for two minutes. “In the control group, air appeared in the coronary arteries within a few seconds after atriotomy and produced arterial obstruction, and evidence of coronary insufficiency appeared. Once the air embolus ceased to move, it remained permanently in that position.” “The heart became dilated and the cardiac action became so weak that cardiac massage was necessary to maintain the circulation in 70% of the control group.” In the treatment group, in which the cardiothoracic wound cavity was insufflated with CO2, “the bubbles of carbon dioxide gas would rapidly traverse the extent of the artery and disappear. The cardiac action was satisfactory and cardiac massage was not necessary except in one case.” Ventricular fibrillation occurred in 37.5% of the control group but in none of the CO2 group. Electrocardiographic changes were seen in 87.5% of the control group versus 28.6% in the treatment group. Gross and microscopic examination revealed positive (pathologic) findings in 75% in the control group versus 28.6% in the CO2 group.

Of particular interest were the measures taken to achieve a high content of CO2 in the open cardiothoracic wound. “In the experimental animals carbon dioxide gas was flooded across the operative field at the flow rate of 5 L per minute in a closed room, since the concentration of carbon dioxide in the chest cavity decreases when the wind blows in.” However, the insufflation device was not described. Only with these measures were they able to achieve a CO2 content of between approximately 80% and 90%. However, we are not aware of any clinical study in which insufflation of the cardiothoracic wound with CO2 through a traditional tube reduced morbidity. In fact, a study by Martens et al. (9) did not find a difference in neuropsychological outcome when CO2 was supplied with a tube compared with a control group without CO2 in open-heart surgery. They concluded, “For effective reduction of cerebral and coronary artery emboli, higher levels of CO2 must be achieved in the operating field by more sophisticated means of application.”

The degree of air displacement in a cardiothoracic wound insufflated with CO2 can be assessed by estimating the content of the remaining air (8,11). We estimated this by measuring the O2 concentration, which is approximately one-fifth (20.95%) of air near sea level (10). Air is a mixture of several gases, including O2, but acts as one gas at normal pressure and temperature. Avogadro’s law states that “equal volumes of different gases contain equal amounts of gas molecules at constant pressure and temperature.” This implies that for every five CO2 molecules supplied to a wound cavity, five air molecules are displaced, and approximately one of these five is an O2 molecule. The O2 instrument with a heated ceramic sensor can assess air displacement more accurately and faster than commonly used CO2 sensors that utilize an optical infrared sensor technique to measure 0%–100% CO2. The O2 sensor’s accuracy was 1% of the measured value in the range of 0.0001%–100% O2, which means that the accuracy increases when the O2 and the air content decrease. Moreover, the O2 sensor requires only a 2-mL gas sample volume and has a response time of less than two seconds. The smaller the gas sample needed, the shorter the response time and the less gas measurements interfere with CO2 de-airing. In contrast, CO2 sensors using the infrared technique usually have a constant accuracy of approximately ±2% units of CO2 over the entire range of measurement of 0%–100% CO2, a larger required sampling volume, and a longer response time, usually >10 seconds. Gases are volatile, and their flow pattern is often turbulent. The quick response modality of the O2 sensor enabled us to detect rapid variations of air content over time. Thus, we consider the O2 sensor more suitable for the evaluation of low air contents during steady-state and during changes than optical infrared CO2 sensors.

The air content was measured at the upper level of the right atrium. Because CO2 is heavier than air, we did not expect larger air contents deeper in the cavity. Because only CO2 flows of ≤10 L/min have been reported (2,8,9,11–13), we did not study CO2 flows greater than 10 L/min. Moreover, already at 5 L/min, almost complete air displacement (≤0.65% air) was obtained when the gas-diffuser was used.

Measurements were performed in an anatomical torso model of adult patients on the same operating table in the same fully ventilated operating room as part of our efforts to reproduce the conditions existing in practice as carefully as possible. We therefore feel justified in assuming that the experimental design enabled us to perform a controlled and standardized study of air displacement with CO2 insufflation in a realistic clinical setup. The clinical consequences of efficient CO2 de-airing were not approached in this study. We are currently conducting a randomized clinical trial evaluating the effect of efficient de-airing with the gas-diffuser during open-heart surgery on neuropsychological outcome.

Open-ended tubes with small inner diameters have been described previously for CO2 insufflation of the cardiothoracic wound cavity (9,13). This study revealed a significant difference in efficiency between the 2.5-mm open-ended tube and the gas-diffuser at all CO2 flows studied (Fig. 3). CO2 insufflation with the 2.5-mm tube resulted in median air contents between 82% and 88% at the studied flows. A thin tube’s apparent failure to displace air with CO2 is probably due to turbulence induced by the CO2 jet. The same phenomenon occurs by analogy when filling a pail with water by using a garden hose. At fast flow rates, most of the water splashes out of the pail. By contrast, the pail is quickly filled if the hose is provided with a multiperforated nozzle, resulting in a reduced flow velocity. The gas-diffuser produced very low levels of air (≤0.65%) in the cardiothoracic wound cavity at a CO2 flow of 5 L/min. At faster CO2 flows, the air content decreased further and reached values as low as ≤0.29% at 10 L/min, both with and without an open left pleural cavity, indicating minimal turbulence despite a high CO2 flow.

Earlier studies have considered CO2 de-airing of a wound cavity only during steady-state. This may partly be due to deficiencies of previously used mea-surement techniques. Our measuring technique has made it possible to study the rapid changes in the degree of de-airing that occurs as a result of changes in CO2 flow and the opening of the left pleural cavity. As seen in Figure 3, a CO2 flow of ≥5 L/min was required during continuous CO2 insufflation with the gas-diffuser to reach very low levels of air in the wound cavity model. Continuous insufflation at ≥5 L/min seems to be needed to compensate for the continuous loss of CO2 due to diffusion as well as to the convective air currents around the wound cavity caused by the ventilation system. Figure 5 further illustrates this phenomenon. When CO2 insufflation is discontinued, the air content in the wound cavity rapidly increases. This implies that a single filling of the cardiothoracic wound cavity with CO2 is insufficient and that intermittent periods without CO2 insufflation should be avoided. Because CO2 is not a liquid that remains in the wound but a gas that, although heavy, disperses into the surrounding air, CO2 insufflation has to be continued as long as heart and vessels are open.

Air displacement may also be impaired by the accidental opening of the left pleural cavity. During cardiopulmonary bypass, the lungs are as a rule not ventilated and are allowed to collapse. Thus, opening the pleural cavity will markedly increase the wound cavity’s volume. Such an increase was found not to have a clinically important influence on the wound cavity’s air content during steady-state, although the differences were statistically significant, nor did opening the pleural cavity have any effect when CO2 insufflation was discontinued. It did, however, substantially delay the initial filling of the cavity with CO2.

The wound cavity should be insufflated with CO2 when there is a risk of air entering the circulatory system. If air enters the heart and great vessels at any time during surgery, air will be trapped (14), and air emboli will eventually occur (15,16). CO2 insufflation into the cardiothoracic wound cavity should be started at a flow of 10 L/min at least one minute before the incision of the heart and great vessels, so that potential gas traps will be filled with CO2 instead of air. An open pleural cavity will prolong complete filling of the cardiothoracic wound cavity with CO2. This can be counteracted by prolonging the initial CO2 insufflation or by increasing the CO2 flow. CO2 insufflation should be continued at a CO2 flow of at least 5 L/min until surgical closure of the heart, the great vessels, and the pulmonary veins, to avoid new possible air trapping or air foam formation. This might be relevant not only for open-heart and aortic surgery, but also for conventional coronary artery bypass surgery, if the single-clamp technique is used. The gas-diffuser could be used for efficient de-airing of the standard cardiothoracic wound cavity, including a complete sternotomy. Finally, the efficiency of the gas-diffuser for de-airing a cardiothoracic wound is significantly superior to that of an open-ended tube. Continuing clinical studies will determine whether this difference is of clinical importance. A suitable position for the diffuser is at a depth of approximately 5 cm adjacent to the diaphragm at the caudal part of the cardiothoracic wound, where there is minimal, if any, surgical activity.


This study showed that the gas-diffuser produced efficient de-airing of a cardiothoracic wound model at CO2 flows of ≥5 L/min, whereas an open-ended tube did not achieve this. An open pleural cavity delayed initial CO2 de-airing and did not have a clinically significant influence on air displacement during steady-state. CO2 insufflation of the cardiothoracic wound cavity should be started at a CO2 flow of 10 L/min at least one minute before the incision of the heart, the great vessels, and the pulmonary veins and should be continued at a CO2 flow of at least 5 L/min until surgical closure.

All authors have contributed equally to this work.


1. Eguchi S, Bosher LH Jr. Myocardial dysfunction resulting from coronary air embolism. Surgery 1962; 51: 103–11.
2. Eguchi S, Sakurai Y, Yamaguchi A. The use of carbon dioxide gas to prevent air embolism during open heart surgery. Acta Med Biol 1963; 11: 1–13.
3. Goldfarb D, Bahnson HT. Early and late effects on the heart of small amounts of air in the coronary circulation. J Thorac Cardiovasc Surg 1963; 46: 368–78.
4. Kunkler A, King H. Comparison of air, oxygen and carbon dioxide embolization. Ann Surg 1959; 149: 95–9.
5. Moore RM, Braselton CW Jr. Injection of air and of carbon dioxide into a pulmonary vein. Ann Surg 1940; 112: 212–8.
6. Spencer FC, Rossi NP, Yu S-C, Koepke JA. The significance of air embolism during cardiopulmonary bypass. J Thorac Cardiovasc Surg 1965; 49: 615–34.
7. Mitz MA. CO2 biodynamics: a new concept of cellular control. J Theor Biol 1979; 80: 537–51.
8. Ng SW, Rosen M. Carbon dioxide in the prevention of air embolism during open-heart surgery. Thorax 1968; 23: 194–6.
9. Martens S, Dietrich M, Wals S, et al. Conventional carbon dioxide application does not reduce cerebral or myocardial damage in open heart surgery. Ann Thorac Surg 2001; 72: 1940–4.
10. Lindsay B. Mechanics. In: Gray DE, ed. American Institute of Physics handbook. 3rd ed. New York: McGraw-Hill, 1972: 134.
11. Selman MW, McAlpine WA, Albregt H, Tatan R. An effective method of replacing air in the chest with CO2 during open-heart surgery. J Thorac Cardiovasc Surg 1967; 53: 618–22.
12. Nadolny EM, Svensson LG. Carbon dioxide field flooding techniques for open heart surgery: monitoring and minimizing potential adverse effects. Perfusion 2000; 15: 151–3.
13. Olinger GN. Carbon dioxide displacement of left heart chambers. J Thorac Cardiovasc Surg 1995; 109: 187–8.
14. Orihashi K, Matsuura Y, Hamanaka Y, et al. Retained intracardiac air in open heart operations examined by transesophageal echocardiography. Ann Thorac Surg 1993; 55: 1467–71.
15. Tingleff J, Joyce FS, Pettersson G. Intraoperative echocardiographic study of air embolism during cardiac operations. Ann Thorac Surg 1995; 60: 673–7.
16. van der Linden J, Casimir-Ahn H. When do cerebral emboli appear during open heart operations? A transcranial Doppler study. Ann Thorac Surg 1991; 51: 237–41.
© 2003 International Anesthesia Research Society