Article Text

Download PDFPDF

NORCCAP (Norwegian colorectal cancer prevention): a randomised trial to assess the safety and efficacy of carbon dioxide versus air insufflation in colonoscopy
  1. M Bretthauer1,
  2. E Thiis-Evensen1,
  3. G Huppertz-Hauss1,
  4. L Gisselsson2,
  5. T Grotmol3,
  6. E Skovlund4,
  7. G Hoff1
  1. 1NORCCAP-Centre, Telemark Central Hospital, Porsgrunn, Norway
  2. 2Department of Anaesthesiology, Telemark Central Hospital, Porsgrunn, Norway
  3. 3The Cancer Registry of Norway, Oslo, Norway
  4. 4Section of Medical Statistics, University of Oslo, Oslo, Norway
  1. Correspondence to:
    Dr M Bretthauer, Norwegian Colorectal Cancer Prevention (NORCCAP), Telemark Central Hospital, N-3906 Porsgrunn, Norway;


Background: To eliminate the risk of combustion during electrosurgical procedures and to reduce patient discomfort, carbon dioxide (CO2) insufflation has been recommended during colonoscopy. However, air insufflation is still the standard method, perhaps due to the lack of suitable equipment and shortage of randomised studies.

Aims: This randomised controlled trial was conducted to assess patient tolerance and safety when using CO2 insufflation during colonoscopy.

Patients: Over an eight month period a successive series of patients referred for a baseline colonoscopy due to findings in a flexible sigmoidoscopy screening trial were randomly assigned to the use of either air or CO2 insufflation during colonoscopy.

Methods: End tidal CO2 (ETCO2), a non-invasive parameter of arterial pCO2, was registered before and repeatedly during and after the examination. The patient's experience of pain during and after the examination was registered using a visual analogue scale (VAS). Sedation was not used routinely.

Results: CO2 insufflation was used in 121 patients (51%) and air in 119 patients (49%). The groups were similar in age, sex, and caecal intubation rate. No rise in ETCO2 was registered. There were statistically significant differences in VAS scores between the groups with less pain reported when using CO2.

Conclusions: This randomised study of unsedated patients shows that CO2 insufflation is safe during colonoscopy with no rise in ETCO2 level. CO2 was found to be superior to air in terms of pain experienced after the examination.

  • colonoscopy
  • air
  • carbon dioxide
  • end tidal carbon dioxide
  • ETCO2, end-tidal carbon dioxide
  • FS, flexible sigmoidoscopy
  • NORCCAP, NORwegian Colorectal CAncer Prevention
  • NYHA, New York Heart Association
  • VAS, visual analogue scale

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

In 1952, the American Journal of Surgery focused on the potential risk of explosion in the rectum during electrosurgical polypectomy. Three cases of explosions were presented, suggested to be caused by the presence of intraluminal explosive gas mixtures.1 One year later, Becker was the first to recommend using carbon dioxide (CO2) insufflation to eliminate the risk of gas explosion during electrocoagulation in the colon.2

In 1986, Christopher B Williams described the superiority of CO2 in reducing the risk of gas combustion, its rapid absorption from the colon allowing double contrast barium enema examination to be performed on the same day if required, and the minimal interference of CO2 with colonic blood flow, reducing the risk of ischaemia.3 It has also been claimed that CO2 insufflation during colonoscopy could reduce pain experienced by patients during and after the procedure.3

Nevertheless, air insufflation has remained the standard method in most centres around the world. Until recently, this was partly due to lack of CO2 insufflators adequately tailored for colonoscopy. To our knowledge however there is a striking absence of randomised trials comparing CO2 and air in colonoscopy.

The use of CO2 in colonoscopy could interfere with the body's acid-base balance, as shown in laparoscopic surgery where a rise in arterial pCO2 is frequently observed after CO2 insufflation.4,5 It is therefore also important to assess the safety of CO2 insufflation during colonoscopy.

As recently described, colonoscopy is increasingly considered to be the future approach in colorectal cancer screening.6 It is therefore of great importance to develop methods that can reduce patient discomfort and thus contribute to better public acceptance for colonoscopy.

This trial compared CO2 with air insufflation in colonoscopy with regard to patient pain during and after the examination and investigated whether CO2 insufflation leads to a rise in body CO2 level.



The NORwegian Colorectal CAncer Prevention (NORCCAP) study is an ongoing screening trial for the prevention of colorectal cancer. Fourteen thousand presumptively healthy men and women, aged 55–64 years, living in two separate areas in Norway are randomly drawn from the population registry and invited to undergo a screening flexible sigmoidoscopy (FS) for colorectal adenomas and cancer. Patients with former colonic resections, severe heart or lung disease (New York Heart Association (NYHA) III-IV), or ongoing treatment for malignant disease are excluded. Bioptically verified adenoma at screening FS, irrespective of size, qualifies for a baseline colonoscopy with polypectomy, to be performed within 6–8 weeks after FS.

NORCCAP participants referred for colonoscopy between October 1999 and April 2000 (267 patients) were included in the present study. All examinations were performed by one of three experienced endoscopists. According to recently published guidelines,7 colonoscopies were performed without any routine use of sedation. However, on demand medication with intravenous midazolam was given if indicated, as judged by the endoscopist. For bowel cleansing, a 4 litre polyethylene glycol solution was used, taken orally on the day before the examination.


Participants referred for colonoscopy were successively assigned to appointments as referrals were received and no sessions were available other than for participants in the NORCCAP screening study. Whole day sessions for colonoscopy were randomised for CO2 or air insufflation, using sealed envelopes. Randomisation of whole sessions rather than individual patients was done to avoid unblinding by change of gas couplings between patients. Both participants and endoscopists were blinded with regard to which gas was being used.

Endoscopic examination

Procedures were performed using Olympus video colonoscopes (Olympus, Hamburg, Germany). However, the standard gas/water valves of the endoscopes which only redirect a continuous gas flow into the gut lumen or atmospheric air, were replaced with another type of valve (MIJ-521, Olympus) preventing gas leakage into the environment. To administer gas into the colon, the valve button had to be pushed halfway down. CO2 or air was administered using two different pressure and flow controlled devices connected to the CO2 and air reservoirs provided (Endoscopic CO2 Regulator; Key Med Ltd (Southend-on-Sea, Essex, UK) for CO2, Norsk Hydro Ltd (Oslo, Norway) for air). The endoscopy assistant was responsible for switching on and off the CO2 and air devices, respectively. To prevent unblinding, the devices were placed behind the endoscopy rack and hidden from the view of the endoscopist.

End-tidal CO2 measurements

End-tidal (ET) CO2 has been shown to give adequate approximations of arterial pCO2 in spontaneously breathing adults and is therefore a good and commonly used non-invasive method of expressing arterial pCO2.8

However, before starting the trial we conducted ETCO2 measurements and arterial pCO2 samples simultaneously on five consecutive patients. The two methods were comparable, with a maximum deviation of the ETCO2 value of only 0.2 kPa. Additionally, we performed repeated measurements within the group of investigators to test the validity of the method, with no evidence of method failure.

ETCO2 was measured successively: (1) at the start of each examination, (2) when the endoscope had reached the caecum, (3) when the rectum was passed during withdrawal, and (4) 10 minutes after finishing the examination. At these measuring points, patients were asked to take a deep breath and to expire deeply and slowly through the mouthpiece of the provided mainstream infrared capnograph (Novametrix Ltd, Wallingford, Connecticutt, USA). The endoscopy assistant performed the measurements and registered the readings. Both the participant and endoscopist were blinded to the results.

Measurement of pain

Pain was registered on a questionnaire given to participants immediately after the examination. A 100 mm visual analogue scale (VAS) was used, ranging from “no pain” on the left to “pain as bad as it could be” on the right end. Participants were asked to score the amount of pain experienced at one, three, six, and 24 hours after the examination. In addition, the questionnaire contained a similar VAS scale for the amount of pain experienced during the examination. Once completed, questionnaires were returned by mail to the screening centre.

Statistical analysis

A pilot study was conducted to estimate the SDs of the pain and ETCO2 measurements, respectively.

Regarding the pain measurements, with an assumed SD of 30 mm, we estimated that 240 patients were needed to achieve at least 95% power to detect a 15 mm difference in VAS between the two groups, which was considered to be clinically important. Assuming that the difference in ETCO2 would have to be >0.5 kPa to be clinically important (SD 0.6), the power to detect this with 150 patients was 95%.

For statistical analysis of repeated measurements of pain and ETCO2, ANOVA for repeated measures was used. Some variables were not normally distributed and thus the Wilcoxon rank sum test was used as a supplementary analysis to compare groups at each time point. The proportion of individuals reporting no pain on the VAS was compared at each time point using the χ2 test. Statistical significance was defined as p≤0.05. Only two sided tests were used. Statistical analyses were performed using SPSS 9.0.


The regional ethics committee approved the study protocol. Informed consent was obtained from all participants before entering the trial.


A total of 267 patients were randomised and examined; 249 patients (93%) completed the questionnaire. Ten patients (seven in the air group and three in the CO2 group; p<0.01) received sedation and were excluded from further analysis. Thus 240 patients were included in the study; 121 (51%) were examined using CO2 insufflation and air was used in 119 patients (49%). There were no differences in baseline characteristics between the two groups (table 1). No statistically significant differences were observed between endoscopists. There was a trend towards more rapid caecal intubation in patients in the CO2 group (table 1).

Table 1

Participant and examination characteristics in the CO2 and air groups

The only complication registered was one perforation requiring colonic resection (air group). This occurred after snare polypectomy of a large sessile adenoma with severe dysplasia in the sigmoid colon.

End-tidal CO2

According to power estimates it was considered sufficient to measure ETCO2 in a limited number of patients and hence ETCO2 measurements were restricted to the first 156 examinations performed without sedation, including 81 patients in the air group and 75 patients in the CO2 group. As shown in fig 1, there was no rise in ETCO2 during or after the examination in any group. On the contrary, we observed a significant reduction in ETCO2 levels during examination in both groups (p<0.001). This reduction was more pronounced when air was used. The time point differences between the groups reached significance only for ETCO2 readings registered towards the end of the examination (p=0.01) (fig 1). However, this difference was estimated to be 0.28 kPa (95% confidence interval (CI) 0.06–0.49) and thus far below the limit of clinical relevance (0.5 kPa). Non-parametric analyses at each time point supported these findings.

Figure 1

Mean (SEM) end-tidal CO2 values at the various observation points in the CO2 (n=75) and air (n=81) groups. **p=0.01 compared with the CO2 group at the corresponding time point by repeated measures ANOVA with multiple comparisons.

Pain during and after examination

Figure 2 shows mean pain scores in each group during and after the examination. There were statistically significant differences in pain scores, favouring CO2 insufflation at all observed time points after examination. The overall mean difference was 7.8 mm (95% CI 4.4–11.2) (p<0.001). The pain reduction after examination was significantly more rapid in the CO2 group (p=0.003). The maximum difference (14 mm (95% CI 9–19); p<0.001) was observed one hour after the examination. Comparison of the two groups by non-parametric analysis at each separate time point gave results similar to the overall analysis.

Figure 2

Mean (SEM) visual analogue scale (VAS) scores at the various observation points during and after examination in the CO2 (n=121) and air (n=119) groups. *p<0.05, ***p<0.001 compared with the CO2 group at corresponding time points by the Wilcoxon rank sum test.

An alternative visualisation of the pain score results is to compare the proportion of patients with score zero (no pain) on the VAS. Figure 3 shows these proportions at each measurement. The finding of a clinically relevant difference in pain, favouring the use of CO2, was supported.

Figure 3

Percentage of patients in the CO2 (n=121) and air (n=119) groups who scored 0 (no pain) on the visual analogue scale (VAS) at the observation points during and after examination. Values are mean (SD). ***p<0.001 compared with the CO2 group at corresponding time points by the χ2 test.


This randomised double blind trial in unsedated patients showed that CO2 insufflation during colonoscopy reduced the amount of pain during and after the examination. No rise in ETCO2 during or after the examination was registered.

The caecum reach ratio in this trial (90%) was somewhat lower than the rates probably expected in colonoscopy performed by experienced endoscopists.7 It must be pointed out that the participants in the present trial were different from patients attending a normal hospital endoscopy unit. They were asymptomatic, with adenomas discovered at screening FS, most of a size not even qualifying for colonoscopy in other FS screening trials.9 In addition, any spread of information in the community about painful colonoscopies would probably influence compliance for both screening and later surveillance. The endoscopists in this trial would therefore be more inclined to discontinue the examination if pain was inflicted during the procedure.

Pain during and after examination

Although a VAS scale is considered to be a reliable method for assessment of patient's pain,10 it has been claimed that more patients fail to score on a VAS scale compared with other pain registration methods.11 However, in the present study 93% of the questionnaires were returned, all completed.

The observed mean difference between the CO2 and air groups regarding pain perception was less than the predefined 15 mm on the VAS scale considered to be clinically important (fig 2). However, in the CO2 group, more than 90% of patients reported that they were completely free from pain after the examination whereas in the air group more than 40% of patients reported pain during the first hours after the procedure (fig 3). In our opinion, this difference is large enough to be clinically relevant and shows clearly the superiority of CO2 compared with air regarding patient pain.

As far as we aware, only one study has been published comparing pain perception using CO2 and air insufflation in colonoscopy.12 The authors reported statistically significant reductions in the amount of pain in favour of CO2 at both six and 24 hours after the examination. In their study, all patients received analgesia (meperidine) and a sedative (diazepam) prior to and during the examination. As diazepam has a long lasting effect with a half life of more than 24 hours, sedation amnesia may influence the validity of scores given.13 In our trial, patients receiving sedation were excluded from analysis; hence the VAS scores and ETCO2 measurements were not influenced by sedation. However, the results of the two studies are similar with a reduction in pain using CO2.

In the present study the use of sedation differed between the two groups, with more individuals in the air group requiring sedation. If these patients had not been excluded from the present analyses, the observed differences would have been somewhat larger. Although the need for on demand administration of sedation in this study was judged subjectively by the endoscopist, the endoscopist's observation of the patient's need was consistent with the differences between the groups in VAS scores given by the patients.

End-tidal CO2

The ideal gas for insufflation during colonoscopy should be inert. However, this gas has yet to be found. CO2 is not ideal as it interferes with normal metabolic processes. It has been known for a long time that intraperitoneal CO2 insufflation during laparoscopic surgery causes a rise in arterial pCO2 levels4,5 but this side effect is probably less marked in retroperitoneal compared with intraperitoneal insufflation.14 Furthermore, a laparoscopic procedure is quite different from colonoscopy (mechanical ventilation, Trendelenburg position, CO2 kept under a positive pressure).3 Therefore, there was a need to investigate the effect on pCO2 during insufflation of CO2 in colonoscopy. To measure arterial pCO2, arterial blood samples are needed. This was considered impractical in the present study. We used ETCO2 measurement as an approximation of arterial pCO2. Continuous measurement of ETCO2, often through a nasal or nasopharyngeal canula, has been the preferred method in other studies.8,15 In the present study patients were awake, non-sedated, and would probably not have tolerated such canulae. We therefore measured ETCO2 repeatedly using a mouthpiece connected to a capnograph. This method may not be as accurate as those published previously but it is an easily performed approximation to arterial pCO2 in this setting. The purpose of our measurements was to exclude a clinically significant rise in body CO2 level, not to evaluate exact values within the reference area.

We did not detect any rise in ETCO2 levels. On the contrary, we observed an overall decrease in ETCO2 during and after the procedure, slightly more marked in the air group. A possible explanation for this is that patients may have hyperventilated during the procedure, and that insufflation of CO2 to some extent outweighed this reduction in ETCO2. Rogers reported a rise, although not statistically significant, in arterial pCO2 during colonoscopy.16 In that study, all patients were sedated and sedation is known to cause changes in pCO2 and other metabolic parameters.17 Hence these results are difficult to compare with ours.

No patient with severe heart or lung diseases (NYHA III-IV) was included in this study. Therefore, our results cannot be generalised to these patient categories. The safety for these patients and for sedated patients has to be investigated further before extending recommendations to other than our described patient population.


This randomised trial of unsedated patients showed that CO2 insufflation during colonoscopy is safe with no rise in ETCO2 levels. CO2 was found to be superior to air regarding pain after the examination. In this study, CO2 insufflation led to an almost complete absence of post examination pain. We recommend CO2 insufflation in colonoscopy.


We thank the nurses at the NORCCAP centre without whose enthusiastic support this study would not have been possible. The trial was supported financially by the Norwegian Department of Health and Social Affairs, the Norwegian Cancer Society, the Norwegian Gastroenterological Association, and Endotronic Ltd. The study was presented in part at the UEGW 2000 in Brussels.