Comparison between Carbon Dioxide and Air Insufflation in Colonoscopy : A Systematic Review and Meta-Analysis Based On Randomized Control Trials

Objectives: The colorectal cancer is a common and lethal neoplasia. Colonoscopy detects diseases in the initial stages decreasing the mortality. Pain and abdominal discomfort are usual complaints associated mainly with the use of air insufflation. Carbon dioxide (CO2) is increasingly utilized to augment tolerance and disposition to repeat the examination. Compare which insufflation method is related to less unpleasant symptoms, safer examination and best performance are objectives of the study. Methods: Electronic databases were accessed selecting only randomized controlled trials comparing insufflation with CO2 and ambient air in colonoscopy. The evaluated outcomes were pain, abdominal distension and flatulence, cecal intubation rate, cecal intubation and total procedure time, volume of gas, CO2 measurement, and need of sedation or analgesia, and polyp detection rate. Results: Thirty randomized controlled trials were selected (4854 patients). Meta-analysis showed reduction in pain risk in the CO2 group immediately after the colonoscopy (Risk difference-RD 0.11[0.03, 0.19]), 1h (RD 0.29 [0.24, 0.34]), 3h (RD 0.22[0.11, 0.34]) and 6h (RD 0.21 [0.17, 0.26]) after colonoscopy. The reduction of flatulence risk 1h and 6h after the procedure was greater in CO2 group (RD 0.54 [0.43, 0.66] and RD 0.65[0.38,0.92], respectively). There were no significant differences between the two groups regarding pain during the procedure, pain and flatulence 24h after colonoscopy, abdominal bloating, request for medication, safety, gas volume, polyp detection rate, cecal intubation rate, time to cecum and total procedure time. Conclusions: CO2 insufflation improves tolerance to colonoscopy, reducing pain and flatulence out to 6 hours following the procedure.


Introduction
The colorectal cancer (CRC) is the third most common neoplasia around the world, occupying the fourth position in mortality, with an incidence of 1,360,602 new cases and 693,933 deaths in 2012 according to the World Health Organization (WHO) -Globocan database [1].Colonoscopy is one of the methods indicated for screening of CRC with the objective of detecting the disease in the initial stages, decreasing the mortality [2,3].
However, because of the necessity of gas insufflation for the adequate visualization of the colonic mucosa, patients usually complain about pain and abdominal discomfort during and after the procedure associated mainly with the use of ambient air, which stays in the intestine for a longer period [4] due to the presence of nitrogen gas.To increase the tolerance and the disposition to repeat the examination, the insufflation of carbon dioxide (CO2) is increasingly utilized.CO2 is rapidly absorbed by the intestinal mucosa and subsequently eliminated by breath, which may lead to less pain, flatulence and distension related to the procedure [5][6][7][8].Comparative analyses between the use of CO2 and ambient air in colonoscopy were shown in two meta-analyses published previously [9,10].
The objective of this systematic review and metaanalysis is to update this knowledge through new studies comparing which insufflation method is related to less unpleasant symptoms, faster and safer examinations, and to add outcomes that were not yet described in the literature.

Literature search
Randomized controlled trials (RCTs) that reported the use of CO2 versus ambient air in colonoscopy published until April 2016 were accessed through the following electronic databases: MEDLINE, SCOPUS, EMBASE, LILACS and CENTRAL (BVS), Cochrane Library, CAPES (Brazil) published theses, in addition to access of bibliographic references ("grey literature") to identify additional articles.Considering the MEDLINE database, the used search strategy was: "Carbon dioxide AND (endoscop* OR colonoscop* OR enteroscop*) AND random*".For the additional databases, we employed the terms: "carbon dioxide", "CO2" and "colonoscopy".

Study Selection
For the study selection, there was no restriction of language, year of publication, patient follow-up time or publication status.After searching the titles and abstracts of the articles selected in the initial search, they were further sorted according the following criteria: 1-Study design: RCT.

4-Comparison: intestinal insufflation with ambient air.
5-Outcomes: the evaluated outcomes were pain, abdominal distension and flatulence related to colonoscopy, cecal intubation rate, cecal intubation time and total procedure time, volume of gas used, CO2 measurement at the end of the procedure, need of sedation or analgesia, and polyp detection rate.

Data Extraction
The data were extracted from the studies by two independent reviewers.In the case of opinion divergence during the process of data extraction and analysis, the doubts were taken to a discussion group in scientific methodology, obtaining a common agreement.

Risk of bias
Some data were used to generate the score of each study according to the JADAD score, which is a tool that qualifies the randomized controlled trial according to the description of randomization, blinding and existence of losses during the trial [11].The prognostic differences between the comparison groups were also evaluated and whether the results of each study were assessed adequately and with sufficient follow-up time.

Statistical analyses
The software RevMan 5 (Review Manager Version 5.3.5 -Cochrane Collaboration, Copyright © 2014) was used to perform the meta-analysis of the outcomes [12].The difference between results was calculated as the risk difference for dichotomous variables with the Cochran-Mantel-Haenszel (CMH) statistical method with 95% confidence interval, and as mean difference for continuous variables, using fixed effect and inverse variance as statistical method with 95% confidence interval.The heterogeneity (I2) between the studies was evaluated and modified whenever possible if >50%, conducting a sensitivity analysis when an outlier was identified through a funnel plot.When the sensitivity analysis did not bring any impact in the heterogeneity reduction, the analysis was not described and we opted for the random effect.Forest and funnel plots were used for graphical expression of the results.

Results
Through the initial search strategy, 1,830 trials were identified and we selected 150 of them to completely evaluate their titles and abstracts.Then 120 studies were excluded by the reasons (Figure 1), which resulted in the selection of 30 RCT that were in accordance with the eligibility criteria, with a total of 4,854 patients: 2,469 in the CO2 insufflation group and 2,385 in the ambient air insufflation group.The main characteristics of the selected trials and the individual risk of bias are shown in table 1.The Jadad score was used for the critical evaluation of the methodological quality of the trials included in this systematic review with meta-analysis, obtaining 23 (77%) trials with values greater than or equal to three, which means low risk of bias.In relation to the description of losses, there was no need to exclude any trial, because all described values were lower than 20%.There was sample power calculation in 17 (57%) trials and the intention-to-treat analysis was conducted in seven (27%).The basic characteristics between the groups compared in each trial were similar in 26 (87%) studies and all used adequate tools to measure the outcomes (Table 1).The indications for colonoscopy were similar between the trials (screening, surveillance and diagnostic).There was no uniformity between the trials in relation to the type of colonic preparation used; types and dosages of sedatives/analgesics during the examination; CO2 insufflation system; patient follow-up time and number of colonoscopists involved in each trial.

Abdominal pain
Twenty-eight randomized controlled trials (RCTs) evaluated the pain related to colonoscopy [4,13,14,16,.However, the present meta-analysis included only trials with complete data and the same score of pain measurement.Whenever the continuous variable analysis (pain score mean/median) was not possible, the data were meta-analyzed using dichotomous variables (pain VS zero pain).The pain was evaluated during and immediately after the colonoscopy, as well as 1h, 3h, 6h and 24h after the procedure.

Endoscopic Ultrasound-Guided Fine Needle Aspiration Accurately Diagnoses Smaller Pancreatic Neuroendocrine Tumors Compared To Computer Tomography-Guided Fine Needle Aspiration
Copyright: © 2017 Pisegna, et al.

Volume of gas during colonoscopy
Two trials evaluated the volume of gas used during the colonoscopy, showing that there was no significant statistical difference when the CO2 insufflation group and the ambient air insufflation group were compared (MD -0.08 [-1.03, 0,86] e I2=44%) [15,41].

Request of medication
There was no difference in relation to the request of medication during the procedure between the CO2 and ambient air insufflation groups in the two trials included in this outcome (RD -0.06 [-0.13, 0]; I2=57%) [13,18].

CO2 measurements after examination
The CO2 measurement (mmHg) after the examination was compared between the groups, showing no difference between the two trials included in this outcome (MD 0.23 [-2.12, 2.58]; I 2 =86%) [24,34]

Discussion
The CO2 use in endoscopic examinations demonstrated that it is an efficient strategy to allow a faster absorption by the intestinal mucosa, causing less unpleasant symptoms related to the procedure [5][6][7][8].
In our systematic review, there was no difference between the CO2 insufflation group and the ambient air insufflation group in relation to pain reported during the examination.However, two meta-analysis [9,10] showed that the CO2 was associated to lower pain scores and lower pain risk during the procedure.This difference may be explained because we performed the sensibility analysis in cases of heterogeneity >50%, splitting the trials between the outcomes "pain during the colonoscopy" and "pain at the end of the colonoscopy", which was not done in the two cited meta-analyses, in addition to the inclusion of new trials.
We demonstrated that the pain risk was lower with the use of CO2 in the analysis of 1h, 3h and 6h after the examination; no clear evidence was found between the groups in the outcome of 24h.Similar results were obtained in the studies cited above, except for the 3h analysis, which was not performed.
The CO2 measurement after the procedure was compared between groups, with no significant statistical difference, which is also shown in the trials included in the systematic review by Wu et al [9].It should be taken into account that the indirect measurements of CO2 through transcutaneous or end tidal monitoring may not be reliable.Arterial blood gases are more adequate, but not acceptable by the patients [42].There was also no difference in relation to cecal intubation.Sajid et al [10] demonstrated that the cecum was reached faster using CO2, but our meta-analysis did not arrive at the same conclusion.
Among the strong points of our systematic review and meta-analysis, we can cite the addition of trials of high methodological quality (77% with Jadad>3), the number of involved patients and the outcomes that were not evaluated in the previous meta-analysis, such as abdominal distention, polyp detection rate, total procedure time, request for analgesia/ sedation during the examination.
Among the limitations of this study, we should punctuate that the high heterogeneity of some outcomes led to the execution of sensibility analysis, modifying some results, which is not possible especially in the outcomes that involve only two studies.Examinations performed in different periods (1992-2015) with particulate clinical practices, different employed methodologies and various forms of outcome measurements are some of the reasons.The presence of analgesia/sedation, type of preparation performed, exam time related to the endoscopist experience and the volume of gas used may relate more with pain, abdominal distention and flatulence than with the type of gas used during the examination.
The cost of a CO2 insufflator varies between 7,000 and 7,400 Euros.The cost of the CO2 gas per colonoscopy is less than 1 euro [43].Yamano et al. state that the total cost of one endoscopy increases about 2.5% with the use of CO2 [41].Thus, the costbenefit relationship between these two insufflation methods must be analyzed in other studies, considering the financial reality of dozens of developing countries.

Figure 1 :
Figure 1: PRISMA flowchart: flow of information through the systematic review

Figure 4 a
Figure 4 a): Pain at 1h post procedure (numerical scale of pain =0).Forest plot and funnel plot: a) before sensibility analysis b) after sensibility analysis

Figure 7 :
Figure 7: Pain at 24h post procedure (numerical scale of pain =0).Forest plot after sensibility analysis

Figure 8 :
Figure 8: Cecal intubation rate.Forest plot and funnel plot

Table 1 :
General characteristics of studies and individual risks of bias (Randomization: + adequated/ -inadequaded/ ?not informed; Losses: YES: losses described/ NO: losses not described ; ITT: intention to treat analysis; Prognostic: YES: similar baseline characteristics between groups/ NO: different baseline characteristics between groups/ ?: comparative analysis absent ; Outcomes: YES: adequately measured/ NO: not adequately measured)