Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

In open heart surgery is there a role for the use of carbon dioxide field flooding techniques to reduce the level of post-operative gaseous emboli?

Three Part Question

In [patients undergoing open heart surgery] can [carbon dioxide field flooding] reduce [gas embolisation]?

Clinical Scenario

You are a Specialist Registrar in cardiothoracic surgery attached to a new unit. Some surgeons are enthusiastic in their use of CO2 field flooding to try to reduce air embolization after cardiac procedures that have involved opening the heart or the aorta. Although you understand that as CO2 is a more soluble gas than air and so it would make sense that it would improve de-airing you wonder whether this has been proven to be the case.

Search Strategy

Medline 1966-Dec 2003 using the OVID interface.
[exp thoracic surgery/OR cardiac surgery/OR cardiac surgical procedures/ OR heart surgery/ OR thoracic surgical procedures] AND [carbon dioxide OR field flooding] AND [embolism/ OR embolus/ OR embolization/ OR embolism/ OR emboli/OR neurological event/ OR cerebrovascular accident/OR stroke] AND Limit to English Language.

Search Outcome

103 papers were found using the above search of which 15 were deemed to be relevant. Eleven were out of scope and 1 was rejected on the basis of poor methodology. Most of the papers out of scope concentrated on technical variations to the technique of carbon dioxide field flooding but did not concentrate on its clinical effect. Three papers that came closest to answering the question we asked were reviewed in full. These are listed in the table.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Martens et al,
62 elective patients assigned to either CO2 (Group I) insufflation to the thoracic cavity n = 31 or to (Group II) control group n = 31. According to the parsonnet risk score, 16 patients in group I (52%) and 10 patients in group II (32%) were categorised as being at either high risk or extremely high risk.Prospective randomized controlled trialMortalityGroup I 3%
Group II 16%
No significant difference

P value not declared
Small study confounded by the mismatch in risk stratification between the 2 groups High mortality rate in group II No statistical values given when data is said to be not significant
Myocardial damageCreatinine kinase-MB was more elevated in the treatment group postoperatively and 24 hours after surgery. (38.0+/-4.1 vs 28.0 +/-2.1, p = 0.02, and 33.0 +/- 2.4 vs 20.6 +/- 2.4, p = 0.01
Neurocognitive functionNo tests revealed a statistically significant result

P value not declared
Webb et al,
Trans-oesophageal echocardiography studies were used to monitor the presence or air bubbles in the heart after open heart operations. CO2 field flooding N=56. De-airing procedures N = 22. All patients had extensive de-airing procedures at the end of the procedureProspective non-randomised studyTrans-oesophageal appearance of air bubbles inside the heart or the aortaAll 22 patients having only de-airing manoeuvres exhibited residual foam inside the heart, aorta or both for at least 30 minutes
Of the 56 patients with CO2 flooding, no foam was observed at the time of cardiac contractions in 34 and in 14 additional cases all bubbles disappeared within the first minute. The remaining 8 patients had complete disappearance in 1 to 24 minutes
Non-randomized No independent validation of observations No statistical test has been performed in the presentation of their data
Martens et al,
A porcine model using diffusion-weighted magnetic resonance imaging as a measure of acute brain infarction. 15 pigs injected with boli of gas via the carotid artery. 1 ml/Kg air (n=5) 1ml/Kg carbon dioxide (n=5) 2ml/Kg carbon dioxide (n=5) Diffusion-weighted magnetic resonance imaging of the brain was performed at 2, 5, 10, 15 and 25 min after embolisationExperimental study in a porcine modelClinical observationsAir (n=5) bradycardic 4, apnoea 5, hyperventilation 0, cardiac arrest 2. Low dose CO2 (n=5) bradycardic 0, apnoea 0, hyperventilation 2, cardiac arrest 0. High dose CO2 (n=5) bradycardic 2, apnoea 1, hyperventilation 1, cardiac arrest 0The gas quantities injected in this animal model exceed realistic amounts of air accidentally embolizing during cardiac surgery Possible contamination with air in both carbon dioxide groups No statistical analysis
Neuroradiologic resultsAir (n=5) Hyper intense Signals Ipsilateral 5, Contralateral 4, persistent > 25 min 5, Low dose CO2 (n=5) Ipsilateral 3, Contra lateral 0, persistent > 25 min 0, High dose CO2 (n=5) Ipsilateral 5, Contra lateral 2, persistent > 25 min 2


Empirically in favour of CO2 field flooding is CO2's solubility in blood in comparison to air and its weight which allows it to settle in the wound cavity. The safety of CO2 has been proven by its injection directly into the aortic root in animal models. Clinically carbon dioxide field flooding was used as early as 1958 [Nichols H, et al.], though it was not adopted universally. The interest now in minimally invasive cardiac procedures where de-airing techniques are more difficult to perform has led to resurgence in the interest in and use of CO2 field flooding. After searching the literature and systematically reviewing the pertinent papers 3 papers were finally included as the best evidence that was available on this topic. Martens et als. (2001) prospective randomised study of CO2 insufflation to the thoracic cavity compared to conventional de-airing techniques found no statistically significant differences between the two groups in terms of mortality or neurocognitive function. Unfortunately the study population was small and the results are confounded by the mismatch in risk stratification between the two groups. Although the mortality rates between the groups were not said to be significant the mortality was lower in the group receiving CO2 (1 death versus 5 deaths in the other group) and the number of high risk patients was higher in this group. In addition, although the differences between groups were not said to be significant in terms of neurocognitive function the percentage of patients with a decline in performance was bigger in the group not receiving CO2 (16% v 29%). Creatinine Kinase-MB, however, was more elevated in the CO2 field flooding group postoperatively and 24 hours after surgery. This group was a higher risk group though according to the parsonnet score and this is the likely explanation for the increased levels of Creatinine Kinase MB. Unfortunately the study groups were definitely too small to reveal differences in mortality or in major neurologic adverse events. Webb et als. prospective non randomised trial of CO2 field flooding versus normal de-airing techniques in patients undergoing valve surgery found that patients who had not had field flooding had persistent air bubbles for at least 30 minutes and usually for 45 minutes whereas the group who did have CO2 field flooding had no air bubbles remaining in less than one minute in 48 out of 56 patients. The patients in this study were not randomized to one technique or the other and the trans-oesophageal echocardiography observer was not blinded as to which technique had been used. Martens et als. (2004) porcine model of gas embolus and its effects on cerebral damage involves sophisticated imaging techniques and clinical observations to detect evidence of brain ischaemia induced by emboli of air or CO2. No statistical analysis is attempted within the paper though it is apparent that animals injected with air emboli had more evidence of neuroradiological brain damage and adverse clinical events. The study shows that air emboli are more detrimental then CO2 emboli, however, this is an unrealistic model of the embolic phenomena which occur during cardiac surgery with unrealistically large volumes of gasses being injected. The use of carbon dioxide is not completely innocuous in every case. Several case reports [O'Connor et al] and [Lippmann M. ] have mentioned the elevated blood levels of CO2 which can be reached with field flooding techniques. The high blood CO2 levels can be accompanied by a marked acidosis. Use of CO2 insufflation into the surgical field should be matched with a cautious use of cardiotomy suckers as there can be an accumulation of the gas in the venous reservoir where it can be absorbed into the blood. Webb et al measured perfusion circuit CO2 tensions and found that they were usually within normal limits, most often ranging from 36 to 50 mmHg and did not cause significant acid-base disturbance. Increased carbon dioxide tension levels correlated with intensive intrapericardial suctioning and quickly returned to normal when suctioning was discontinued. Martens et al (2002) conducted a study to determine the optimal technique for the delivery and amounts of CO2 which should be delivered into the operative field. They measured arterial blood gasses and found that they could deliver CO2 at up to 10 L/minute without causing an increase in arterial PCO2 or acidosis.

Clinical Bottom Line

There is no large clinical study to prove that there is a neurocognitive benefit to the use of CO2 field flooding. A single small study which was inadequately powered could not detect a significant benefit to the use of CO2 field flooding but the results pointed towards a superior outcome in the treatment group and it is possible that a larger study would confirm a benefit. Despite this we conclude that the inherent solubility of carbon dioxide emboli justify efforts to replace intracavital air by CO2 in open heart surgery but that caution is warranted as use of excessive cardiotomy suction may result in hypercarbia and a large clinical controlled randomised prospective study has not been conducted to prove that there is any benefit. A large clinical controlled randomised prospective study with risk stratification would be of interest, but would be difficult to undertake.


  1. 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(6):1940-4.
  2. Webb WR, Harrison LH Jr, Helmcke FR, et al. Carbon dioxide field flooding minimizes residual intracardiac air after open heart operations. Ann Thorac Surg 1997;64(5):1489-91.
  3. Martens S, Theisen A, Balzer JO, et al. Improved cerebral protection through replacement of residual intracavital air by carbon dioxide: a porcine model using diffusion-weighted magnetic resonance imaging. J Thorac Cardiovasc Surg 2004;127(1):51-6.
  4. Nichols H, Morse P, Hirose T. Coronary and other air embolization occurring during open cardiac surgery. Surgery 1958;43:236-44.
  5. O'Connor BR, Kussman BD, Park KW. Severe hypercarbia during cardiopulmonary bypass: a complication of CO2 flooding of the surgical field. Anesth Analg 1998;86(2):264-6.
  6. Lippmann M. Complications of CO2 flooding the surgical field in open heart surgery: an old technique revisited. Anesth Analg 1998;87(4):978-9.
  7. Martens S, Dietrich M, Doss M, et al. Optimal carbon dioxide application for organ protection in cardiac surgery. J Thorac Cardiovasc Surg 2002;124(2):387-91.