Water exchange colonoscopy increases adenoma detection rate: a systematic review with network meta-analysis of randomized controlled studies

SYSTEMATIC REVIEW AND META-ANALYSIS

Water exchange colonoscopy increases adenoma detection rate: a systematic review with network meta-analysis of randomized controlled studies Lorenzo Fuccio, MD,1 Leonardo Frazzoni, MD,1 Cesare Hassan, MD, PhD,2 Marina La Marca, MD,1 Valentina Paci, MD,1 Veronica Smania, MD,1 Nicola De Bortoli, MD,3 Franco Bazzoli, MD,1 Alessandro Repici, MD,4 Douglas Rex, MD,5 Sergio Cadoni, MD6 Bologna, Rome, Pisa, Rozzano, Iglesias, Italy; Indianapolis, Indiana, USA

Background and Aims: Water-aided colonoscopy techniques, such as water immersion (WI) and water exchange (WE), have shown different results regarding adenoma detection rate (ADR). We determined the impact of WI and WE on ADR and other procedural outcomes versus gas (air, AI; CO2) insufflation colonoscopy. Methods: A systematic search of multiple databases for randomized controlled trials comparing WI and/or WE with AI and/or CO2 and reporting ADR was conducted. A network meta-analysis with mixed comparisons was performed. Primary outcome was ADR (overall, in the right side of the colon and by colonoscopy indication). Results: Seventeen randomized controlled trials (10,350 patients) were included. WE showed a significantly higher overall ADR versus WI (odds ratio [OR], 1.31; 95% credible interval [CrI], 1.12-1.55) versus AI (OR, 1.40; CrI, 1.221.62) versus CO2 (OR, 1.48; 95% CrI, 1.15-1.86). WE achieved the highest ADR also in the right side of the colon and in colorectal cancer screening cases (both significant vs AI and WI) as well as in patients taking a split-dose preparation (significant vs all the other techniques). The Boston Bowel Preparation Scale cleanliness score (vs AI and WI) was significantly higher for WE. Both WI and WE showed increased proportion of unsedated examinations and decreased real-time insertion pain, with WE being the least-painful insertion technique. Withdrawal time was comparable across techniques, but WE showed the longest insertion time (3-5 additional minutes). Conclusions: WE significantly increases overall ADR, ADR in screening cases, and in the right side of the colon; it also improves colon cleanliness but requires a longer insertion time. (Gastrointest Endosc 2018;-:1-9.)

Water-aided colonoscopy (WAC) encompasses different techniques that entail infusion of water as an adjunct or in lieu of gas insufflation to distend the lumen during the insertion phase.1 It is broadly categorized as water immersion or infusion (WI) and water exchange (WE).1 Numerous randomized controlled trials (RCTs) described WAC and refined what is meant by the terms WI

Abbreviations: ADR, adenoma detection rate; AI, air; BBPS, Boston Bowel Preparation Scale; CI, confidence interval; CrI, credible interval; CRC, colorectal cancer; MD, mean difference; OR, odds ratio; RCT, randomized controlled trial; WAC, water-aided colonoscopy; WE, water exchange; WI, water immersion. DISCLOSURE: D. Rex, consultant for and the research support from Boston Scientific and Olympus. All authors disclosed no financial relationships relevant to this publication. Copyright ª 2018 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 https://doi.org/10.1016/j.gie.2018.06.028 (footnotes continued on last page of article)

www.giejournal.org

and WE.2-11 Accordingly, WE is a standardized technique developed to achieve gasless colonoscope insertion, substituting through infusion and near-simultaneous suction of water all colon content with a layer of clear water, allowing instrument progression to the cecum. This technique minimizes distension-related pain and maximizes colon cleanliness during insertion to increase the ease of withdrawal inspection.1,12 Infused water is removed predominantly during insertion. On the other hand, WI is an unstandardized technique in which water is infused to facilitate cecal intubation.2-6 WI does not entail suction removal of all dirty water, residual air, and feces during insertion and has been deemed too cumbersome and time-consuming.2-4 Indeed, when judged to be safe, colonoscope insertion is done also in opaque water and/or air compartment.2-4,13 Gas insufflation is used in limited amounts when necessary2,5 or is resumed in the proximal colon to achieve cecal intubation.13 Infused water is removed predominantly during withdrawal. With both techniques, withdrawal is usually carried out using gas insufflation. Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 1

WE colonoscopy increases adenoma detection rate

Fuccio et al

Adenoma detection rate (ADR) is an important colonoscopy quality indicator,14 and its increase has been linked to a reduced risk of interval cancer15 and death.16 When analyzed separately, compared with gas insufflation colonoscopy (air [AI]) or CO2, the impact of WI and WE on ADR differs.1 WI achieved inconsistent results,2-6,10 whereas WE has been shown to consistently increase ADR,17-19 also in the right side of the colon.10,20 However, these observations originated from studies with small sample sizes or with adenoma detection as a secondary outcome. Furthermore, either study design21 or sample size9,11 precluded demonstrating a significant difference between WE and WI. Therefore, we conducted a systematic review and network meta-analysis to determine whether WI or WE is superior to AI and/or CO2 in increasing ADR, also in the right side of the colon. We were also interested in the impact of these insertion techniques on colon cleanliness, which influences adenoma detection,22-25 and other colonoscopy procedural outcomes.

METHODS We followed the PRISMA guideline and checklist for reporting systematic reviews and network meta-analyses.26 Data sources and search strategy, selection process, data extraction, and quality assessment are reported in detail in Appendix 1 (available online at www.giejournal.org).

Inclusion and exclusion criteria A literature search was done to identify all relevant RCTs comparing WI and/or WE with AI and/or CO2 insufflation published since 1999 as full text in English and including ADR among the outcomes. EMBASE, Medline, SCOPUS, and Cochrane Library were searched systematically for all articles that included the following terms in their titles, abstracts, or keyword lists: water-aided colonoscopy, water immersion colonoscopy, water exchange colonoscopy, air insufflation colonoscopy, carbon dioxide insufflation colonoscopy, and adenoma detection rate (see Appendix 1). References in retrieved articles were screened manually. Exclusion criteria included reviews, case reports, editorials, commentaries, articles involving only supervised trainees, articles including limited use of WI or WE in the distal colon, articles where WAC was performed with add-ons (devices, oil), and/or chromoendoscopy.

Outcomes assessment The primary outcome was overall ADR27 (defined as ADR from screening, surveillance, and diagnostic procedures) according to the colonoscopy technique and separately assessed in the entire and in the right side of the colon (cecum and ascending). Subgroup analyses in screening patients and by bowel preparation (ie, daybefore or split-dose) were also carried out. 2 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

A key secondary outcome was bowel cleansing according to the Boston Bowel Preparation Scale (BBPS). Other secondary outcomes included cecal intubation time, withdrawal time (cases with and without polypectomies), proportion of complete unsedated procedures, real-time insertion pain, and adverse events. We used Cochrane’s collaboration risk of bias tool to assess the risk of bias in all the included studies.28

Data analysis First, we ran traditional meta-analyses with random effect models considering all studied outcomes for each colonoscopy technique separately to provide their absolute estimates. Data were summarized as pooled estimates of proportions for categorical outcomes and means for continuous outcomes, with 95% confidence interval (CI). Second, we ran traditional pairwise meta-analyses with random effects models for all colonoscopy techniques compared within the included studies calculating the pooled estimates of odds ratio (OR) and 95% CI of direct comparisons between any 2 endoscopic techniques. Statistical heterogeneity was quantified with the I2 statistic (high heterogeneity level >50%) and tested using the Cochrane Q2 test (statistical significance level, P < .1). Publication bias was assessed by the Egger regression asymmetry test. For the Cochrane Q2 test and the Egger test, a 2-sided P < .05 was regarded as significant. Third, we conducted the network meta-analysis using the Bayesian Markov Chain Monte Carlo method. A randomeffects model with noninformative priors comparable was used given the nature of network meta-analysis, encompassing tolerability to between-study heterogeneity and withinstudy variability. Direct and indirect evidence for all colonoscopy techniques were combined to estimate the studied outcomes, with a 95% equal tail credible interval (CrI).29 Results were then presented as relative effects and Bayesian estimates of the probability that each technique has to be the best, the second best, the third best, and the worst relating to every studied outcome. All analyses were done with R software30 by means of gemtc package31 and metafor package.32

RESULTS Our search identified 155 publications. Seventeen RCTs2-6,9-11,17-21,33-36 were included in the analysis (41 arms of treatment, 10,350 patients; 55.7% male patients; Supplementary Figs. 1 and 2, available online at www. giejournal.org). Baseline characteristics of the included studies are reported in Table 1. The raw proportions and means of the studied outcomes, stratified by each colonoscopy technique, are shown in Supplementary Table 2 (available online at www.giejournal.org) to provide their absolute magnitude. The results of multiple comparisons regarding all the www.giejournal.org

Fuccio et al

outcomes are summarized in Table 2 (network metaanalysis) and Supplementary Table 1 (traditional pairwise meta-analysis), available online at www.giejournal.org.

Primary outcome: ADR In 17 trials with 41 study arms and 10,350 patients,2-6,9-11,17-21,33-36 the raw estimates of overall ADR (95% CI) were as follows: 41.7% (95% CI, 32.5%-51.5%) for WE, 34.4% (95% CI, 28.3%-40.9%) for WI, 30.2% (95% CI, 24.4%-36.8%) for AI, and 31.1% (95% CI, 19.0%46.4%) for CO2 insufflation. In the network meta-analysis, WE had a significantly higher entire colon overall ADR when compared with WI (OR, 1.31; 95% CrI, 1.12-1.55), AI (OR, 1.40; 95% CrI, 1.22-1.62), and CO2 insufflation (OR, 1.48; 95% CrI, 1.15-1.86). WI was not superior to AI or CO2 insufflation regarding ADR. ADR in the right side of the colon was 21.1% (95% CI, 16.4%-26.8%) for WE, 15.8% (95% CI, 12.5%-19.8%) for WI, and 14.2% (95% CI, 11.6%-17.3%) for AI, comprising 4 trials with 12 study arms and 3345 patients.9-11,20 According to the network analysis, WE showed a significantly higher right side of the colon ADR than AI (OR, 1.54; 95% CrI, 1.23-1.93) and WI (OR, 1.36; 95% CrI, 1.10-1.70). WI did not differ from AI. According to the Bayesian framework, WE had the highest probability of being rated as the best technique for both the primary outcomes (Fig. 1).

Subgroup analyses of primary outcome Six studies with 14 study arms and 6300 patients evaluated colorectal cancer (CRC) screening ADR in the entire colon.9,11,17-19,21 Enrolled patients were average risk (3 studies), fecal immunochemical test positives (1 study) or both (2 studies). The raw estimates were 42.4% (95% CI, 32.2%53.3%) for WE, 39.5% (95% CI, 30%-49.9%) for WI, and 32.3% (95% CI, 23.7%-42.3%) for AI. In the network analysis, CRC screening ADR was significantly higher for WE than AI (OR, 1.48; 95% CrI, 1.19-1.9) or WI (OR, 1.4; 95% CrI, 1.022.09). WI was not superior to AI. Lack of data hindered us to analyze the right side of the colon ADR in CRC screening. Considering 10 studies with 25 arms and 8716 patients conducted with split-dose bowel preparation, the raw estimates of ADR in the entire colon were 40.7% (95% CI, 29.2%-53.4%) for WE, 37.6% (95% CI, 32.3%-43.2%) for WI, 37.9% (95% CI, 21.3%-58%) for CO2, and 28% (95% CI, 20.6%-36.8%) for AI. In the network meta-analysis, WE obtained significantly higher ADRs than AI (OR, 1.43; 95% CrI, 1.22-1.67), CO2 (OR, 1.39; 95% CrI, 1.05-1.85), and WI (OR, 1.26; 95% CrI, 1.04-1.51). WI did not perform better than AI or CO2. According to the Bayesian framework, WE had the highest probability of being rated as the best technique for all the analyzed outcomes (Fig. 1).

Key secondary outcome: Bowel cleansing according to mean BBPS The pooled estimates of bowel cleansing as evaluated by mean BBPS of the 4 techniques were derived from 6 www.giejournal.org

WE colonoscopy increases adenoma detection rate

studies using split-dose preparation (16 arms, 7121 patients).9-11,20,21,36 BBPS scores were 7.4 (95% CI, 7.07.8) for WE, 7.0 (95% CI, 6.5-7.5) for WI, 6.8 (95% CI, 6.3-7.3) for AI, and 7.3 (95% CI, 7.0-7.6) for CO2 insufflation. In the network analysis, WE performed significantly better than AI (mean difference [MD], .68; 95% CrI, .34-1.02) and WI (MD, .41; 95% CrI, .04-.78). WE and CO2 insufflation were comparable (MD, 0; 95% CrI, –.80 to .82). WI was comparable with both AI and CO2 insufflation.

Other secondary outcomes Considering 17 trials (10,350 patients, 41 study arms),2-6,9-11,17-21,33-36 cecal intubation time in mean minutes was 13.4 (95% CI, 9.1-17.7) for WE, 8.3 (95% CI, 7.4-9.2) for WI, 10.2 (95% CI, 6.6-13.8) for AI, and 8.5 (95% CI, 6.8-10.2) for CO2 insufflation. In the network setting, WE achieved significantly higher cecal intubation time, as compared with AI (MD, 2.62; 95% CrI, .66-4.52), CO2 insufflation (MD, 3.3; 95% CrI, .49-6.07), and WI (MD, 3.32; 95% CrI, 1.03-5.58). WI did not differ from both AI and CO2 insufflation. Withdrawal time was reported by 9 trials (22 study arms, 7007 patients) and was not significantly different between the 4 colonoscopy techniques.6,10,17,19-21,33-35 Considering 12 trials with 27 study arms and 7599 patients,2,3,5,9,17-19,21,33-36 the proportion of nonsedated procedures was 86.7% (95% CI, 52.1%-97.5%) for WE, 83.7% (95% CI, 63.9%-93.7%) for WI, 75.1% (95% CI, 45.5%-91.6%) for AI, and 97.1% (95% CI, 82.8%-99.6%) for CO2 insufflation. In the network meta-analysis, WE and WI achieved significantly higher proportions of nonsedated procedures when compared with AI (OR, 1.65; 95% CrI, 1.1-2.6; and OR, 1.64; 95% CrI, 1.02-2.8, respectively), whereas other comparisons were not significant. Real-time insertion pain was reported in 9 trials with 21 study arms and 5508 patients3,4,10,11,18,21,33,34,36 as expressed by mean numeric rating scale or visual analog scale and was 2.8 (95% CI, 1.7-3.9) for WE, 2.9 (95% CI, 2.3-3.5) for WI, 4.6 (95% CI, 4.0-5.2) for AI, and 4.1 (95% CI, 1.7-6.4) for CO2 insufflation. In the network analysis, real-time insertion pain was significantly higher for AI versus all other techniques. In detail, CO2 versus AI yielded an MD of –1.89 (95% CrI, –3.23 to –.58), WI versus AI yielded an MD of –1.54 (95% CrI, –2.37 to –.72), and WE versus AI resulted in an MD of –1.99 (95% CrI, –2.82 to –1.21). Other comparisons were not significant. Adverse events (41 vagal reactions, 25 bleeding episodes) were reported by 12 trials (29 arms, 8264 patients).2,4-6,9,11,17,21,33-36 Adverse events were comparable among the 4 techniques. Supplementary Figure 3 (available online at www. giejournal.org) shows the Bayesian probability of ranking as the best technique for the secondary outcomes. WE had the highest probability of being the most timeconsuming technique. Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 3

WE colonoscopy increases adenoma detection rate

Fuccio et al

TABLE 1. Baseline characteristics of the included studies Country

Total sample size

Techniques compared

Mean age (y)

Leung CW, 20104

USA

226

AI vs WI

62.5 vs 62.5

114/114 (100%) vs 112/112 (100%)

Amato A, 20135

Italy

341

AI vs CO2 vs WI

60 vs 61.5 vs 60

72/113 (64%) vs 75/115 (65%) vs 44/113 (39%)

Radaelli F, 20102

Italy

230

AI vs WI

58.8 vs 58.4

66/114 (58%) vs 68/116 (59%)

Hsieh YH, 201410

Taiwan

270

AI vs WI vs WE

56.5 vs 54.3 vs 56.9

60/90 (67%) vs 53/90 (59%) vs 55/90 (61%)

Falt P, 20126

Czech Republic

404

AI vs CO2 vs WI

58.7 vs 59.4 vs 59.1

54/101 (53%) vs 50/102 (49%) vs 105/201 (52%)

Pohl J, 20113

Germany

116

AI vs WI

61.7 vs 62.7

42/58 (72%) vs 43/58 (74%)

Xu X, 201634

China

287

AI vs CO2 vs WI

55 vs 54.3 vs 54.3

45/94 (48%) vs 49/96 (51%) vs 46/97 (47%)

Ramirez J, 201119

USA

368

AI vs WE

59.3 vs 60

184/191 (96%) vs 171/177 (97%)

Leung J, 201133

USA

100

AI vs WE

58.3 vs 60.7

50/50 (100%) vs 49/50 (98%)

Arai M, 201635

Japan

403

CO2 vs WE

63.1 vs 65.6

122/197 (62%) vs 125/206 (61%)

Garborg K, 201536

Norway, Poland

473

CO2 vs WE

61 vs 60

121/239 (51%) vs 120/234 (51%)

Cadoni S, 201417

Italy

672

AI vs WE

60 vs 58

204/334 (61%) vs 201/338 (60%)

Leung FW, 201018

USA

82

AI vs WE

66,8 vs 66

0/40 (0%) vs 0/42 (0%)

Italy and Czech Republic

1224

AI vs WI vs WE

60.9 vs 61 vs 61.4

225/408 (55%) vs 223/408 (55%) vs 224/408 (55%)

Hsieh Y-H, 201711

Taiwan

651

AI vs WI vs WE

54.8 vs 55.9 vs 55.7

105/217 (48%) vs 110/217 (51%) vs 121/217 (56%)

Jia H, 201721

China

3303

AI vs WE

50.3 vs 50.3

855/1650 (52%) vs 830/1653 (50%)

Italy, Czech Republic

1200

AI vs WI vs WE

59 vs 60 vs 58

239/401 (60%) vs 233/404 (58%) vs 232/395 (59%)

Study

Cadoni S, 20171

Cadoni S, 201620

Male gender n/N

CRC, Colorectal cancer; ADR, adenoma detection rate; d, not available; AI, air; WI, water immersion; WE, water exchange.

Risk of bias The risk of bias assessment is reported in Appendix 2 (available online at www.giejournal.org).

DISCUSSION The main finding of the current systematic review with network meta-analysis is that WE was associated with the highest ADR (overall, CRC screening cases and in patients 4 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

taking a split-dose preparation) than all other colonoscopy techniques, including WI, with a 97% Bayesian probability of ranking as the most effective. Notably, WE was also associated with an increase in the right side of the colon ADR (93% Bayesian probability of being rated as best technique). The novelty of the current network meta-analysis is the head-to-head assessment between WI and WE, integrating both direct and indirect evidence to compare their effectiveness for adenoma detection and other colonoscopy www.giejournal.org

Fuccio et al

WE colonoscopy increases adenoma detection rate

TABLE 1. Continued CRC screening indication n/N

Overall ADR in entire colon

Overall ADR in right side of colon

ADR in entire colon, CRC screening

ADR in right side of colon, CRC screening

80/114 (70%) vs 71/112 (63%)

45/114 (40%) vs 47/112 (42%)

d

d

d

43/113 (38%) vs 33/115 (29%) vs 37/113 (33%)

42/113 (37%) vs 35/115 (30%) vs 44/113 (39%)

d

d

d

42/114 (37%) vs 46/116 (40%)

46/114 (40%) vs 29/116 (25%)

d

d

d

4/90 (4%) vs 1/90 (1%) vs 3/90 (3%)

39/90 (43%) vs 41/90 (46%) vs 51/90 (57%)

10/90 (11%) vs 13/90 (14%) vs 24/90 (27%)

d

d

25/101 (25%) vs 26/102 (26%) vs 56/201 (28%)

22/101 (22%) vs 27/102 (27%) vs 71/201 (35%)

d

d

d

37/58 (64%) vs 41/58 (71%)

15/58 (26%) vs 19/58 (33%)

d

d

d

d

12/94 (13%) vs 14/96 (15%) vs 13/97 (13%)

d

d

d

191/191 (100%) vs 177/177 (100%)

88/191 (46%) vs 101/177 (57%)

d

88/191 (46%) vs 101/177 (57%)

d

36/50 (72%) vs 36/50 (72%)

18/50 (36%) vs 20/50 (40%)

d

d

d

d

114/197 (58%) vs 139/206 (67%)

d

d

d

219/239 (92%) vs 208/234 (89%)

74/239 (31%) vs 81/234 (35%)

d

d

d

122/334 (37%) vs 111/338 (33%)

64/334 (19%) vs 87/338 (26%)

d

35/122 (29%) vs 44/111 (40%)

d

19/40 (48%) vs 17/42 (41%)

9/40 (23%) vs 15/42 (36%)

d

3/19 (16%) vs 5/42 (12%)

d

408/408 (100%) vs 408/408 (100%) vs 408/408 (100%)

165/408 (40%) vs 177/408 (43%) vs 201/408 (49%)

69/408 (17%) vs 78/408 (19%) vs 98/408 (24%)

165/408 (40%) vs 177/408 (43%) vs 201/408 (49%)

69/408 (17%) vs 78/408 (19%) vs 98/408 (24%)

63/217 (29%) vs 67/217 (31%) vs 55/217 (25%)

82/217 (38%) vs 88/217 (41%) vs 108/217 (50%)

33/217 (15%) vs 38/217 (18%) vs 47/217 (22%)

24/63 (38%) vs 22/67 (33%) vs 28/55 (51%)

d

541/1650 (33%) vs 496/1653 (30%)

221/1650 (13%) vs 303/1653 (18%)

d

111/541 (21%) vs 129/496 (26%)

d

97/401 (24%) vs 101/404 (25%) vs 90/395 (23%)

129/401 (32%) vs 117/404 (29%) vs 131/395 (33%)

48/401 (12%) vs 49/404 (12%) vs 59/395 (15%)

d

d

procedural outcomes. According to our analysis, WE is superior to WI for both overall ADR and the right side of the colon ADR and for ADR in screening setting. Although only 6 studies reported ADR by indication (ie, screening or diagnostic setting), our results on ADR are further strengthened by considering that overall and screening ADR were above the 25% threshold recently proposed as minimum standard for colonoscopy quality.37 The exact cause of the increase in ADR by WE colonoscopy is not fully understood. A first plausible mechanism is www.giejournal.org

the improved quality of cleanliness achieved through continuous infusion and suction of water. In our analysis, the bowel preparation used was not likely to favor any particular technique. As far as overall and screening ADR were concerned, the median number of patients that used day-before and split-dose preparation was comparable across insertion techniques. The turbulences created at the tip of the instrument by the water jet effectively dislodge residual feces and debris adherent to the mucosa, which are then removed by suction.9 Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 5

WE colonoscopy increases adenoma detection rate

Fuccio et al

TABLE 2. Meta-analysis outcomes as obtained by network meta-analysis Primary outcome (OR, 95% CrI)

WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

Overall ADR in entire colon

1.4 (1.22-1.62)

1.48 (1.15-1.86)

1.31 (1.12-1.55)

1.07 (.92-.26)

1.13 (.87-1.45)

.95 (.75-1.23)

Overall ADR in right side of colon

1.54 (1.23-1.93)

d

1.36 (1.10-1.70)

1.13 (.90-1.43)

d

d

ADR in entire colon, CRC screening

1.48 (1.19-1.9)

d

1.4 (1.02-2.09)

1.06 (.72-1.46)

d

d

ADR in entire colon, diagnostic/surveillance

1.60 (1.16-2.24)

d

1.25 (.71-2.22)

1.29 (.72-2.27)

d

d

ADR in entire colon, split-dose preparation

1.43 (1.22-1.67)

1.39 (1.05-1.85)

1.26 (1.04-1.51)

1.13 (.94-1.38)

1.1 (.81-1.53)

1.03 (.75-1.39)

WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

Bowel cleansing, BBPS

.68 (.34-1.02)

0 (–.8 to .82)

.41 (.04 to .78)

.26 (–.1 to .65)

–.41 (–1.29 to .49)

.68 (–.2 to 1.56)

Cecal intubation time, min

2.62 (.66-4.52)

3.3 (.49-6.07)

3.32 (1.03-5.58)

–.7 (–2.61 to 1.18)

–.02 (–2.8 to 2.78)

–.68 (–3.4 to 2.01)

Secondary outcome (mean difference, 95% CrI)

Withdrawal time, min

.1 (–.76 to 1.02)

.29 (–.94 to 1.62)

–.54 (–1.73 to .65)

.64 (–.39 to 1.73)

.83 (–.39 to 2.16)

–.19 (–.43 to 1)

Real-time insertion pain, NRS/VAS

–1.99 (–2.82 to –1.21)

–.1 (–1.43 to 1.2)

–.45 (–1.46 to .51)

–1.54 (–2.37 to –.72)

.35 (–1.02 to 1.73)

–1.89 (–3.23 to –.58)

Secondary outcome (OR, 95% CrI)

WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

1.65 (1.1-2.6)

1.01 (.47-2.18)

1.01 (.55-1.79)

1.64 (1.02-2.8)

1 (.45-2.31)

1.64 (.77-3.59)

.9 (.37-2.12)

2.01 (.3-17.34)

1.46 (.55-3.93)

.61 (.27-1.34)

1.37 (.23-11.53)

.45 (.06-2.51)

Proportion of unsedated procedures Adverse events rate

Values are ORs (categorical variables) or mean difference (continuous variables), along with 95% credible interval (CrI). Significant comparisons are in bold characters. ADR, Adenoma detection rate; BBPS, Boston Bowel Preparation Scale; CI, confidence interval; CRC, colorectal cancer; NRS/VAS, numeric rating scale/visual analog scale; OR, odds ratio; AI, air insufflation; WE, water exchange; WI, water immersion; d, not applicable.

According to our data WE, but not WI, significantly increased BBPS scores, both in the entire and the right side of the colon. Second, the reduced use of washing and suction during the withdrawal phase has been shown to result in fewer collapses of the lumen and wall contractions of the colon; thus, the endoscopist can concentrate on the main task of inspecting the mucosa searching for lesions without distractions.38 Third, during the insertion phase polyps are more evident because the underwater view has a magnifying effect; additionally, lesions are less flattened and float underwater. Thus, a higher ADR can be achieved also during insertion.10 It could be argued that the higher ADR by WE is because of its longer insertion time, with more lesions found during insertion that are relocated during withdrawal. Nevertheless, in a multicenter RCT with ADR as the primary outcome and with colonoscopists blinded to the insertion method,9 WE ADR (49.3%) was higher than AI (40.4%, P Z .03) and WI (43.4%, P Z .28, not significant because of lack of statistical power). Taken together, the above explanations suggest that the improved cleanliness and the lack of distractions during withdrawal contribute toward a higher ADR by WE.

6 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

The correct implementation of WE is critical to achieve an increase in adenoma detection. In a head-to-head comparison of WE, WI, and AI for ADR in the screening setting and with colonoscopists blinded to the insertion method, WE performed with a low volume of water did not achieve significantly higher ADR or BBPS.9 Thus, it may be inferred that salvage cleaning provided by WE during insertion and the volume of instilled water are critical to its success in improving ADR. However, the amount of water to be infused is not standardized, and it should be as much as a thin layer of clear water is obtained for progression. Standardization of the procedures across most of the included studies was achieved because Dr F. W. Leung, who pioneered the use of WE, is listed as co-author or provided hands-on coaching in a substantial subset of the studies. In addition, in all studies gas insufflation was turned off before starting the procedure and not allowed using WE, and its use was considered as an intention-to-treat failure. We speculate that the higher ADR achieved by WE suggests that in a similar span of time more work was done resecting lesions, offsetting the longer time required to reach the cecum.

www.giejournal.org

Fuccio et al

WE colonoscopy increases adenoma detection rate

Bayesian probability of ranking as the best technique 99%

97%

100%

98% 93%

90%

90% 80% 70% 60% 50% 40% 30% 20% 10%

7%

10% 1% 1% 1%

0%

1%

0%

0% 1% 1%

0% 0%

0% ADR in entire colon, all settings

ADR in right colon, all ADR in entire colon, CRC ADR in entire colon, screening setting diagnostic/surveillance settings setting

Al

CO2

WI

ADR in split-dose

preparation

WE

Figure 1. Bayesian probability of ranking as the best technique for adenoma detection rate of the various colonoscopy techniques, as estimated by network meta-analysis. ADR, Adenoma detection rate; CRC, colorectal cancer; AI, air; WI, water immersion; WE, water exchange.

Unfortunately, we were unable to examine withdrawal time of cases with and without polypectomy separately because of the limited information available. The observation that withdrawal time was similar among the 4 techniques suggests that WE can be used in many endoscopy practices at the expense of a few more minutes of insertion time but with the benefit of a significant improvement in adenoma detection. WI and WE imposed fewer burdens on patients in terms of real-time insertion pain and sedation than AI. Lack of significance of WI and WE versus CO2 insufflation for these 2 outcomes was probably because of a paucity of cases available for the analyses. Moreover, because the primary endpoint of the current study was ADR, we included patients regardless of the type of sedation used (on-demand, moderate, or deep). This may have lessened the impact of the pain-alleviating effect of WE compared with WI and/or CO2 insufflation. However, several RCTs have already demonstrated that WE is superior to WI in decreasing real-time insertion pain,7,8,10 and our analysis confirms that WE is the least painful insertion technique. On the contrary, AI was considered the most painful method for the ranking probabilities. The impact of WE on sedation and colonoscopy pain might be relevant in many different parts of the world, where colonoscopies are performed without sedation or using on-demand moderate sedation. One major drawback for the diffusion of WE has been its long insertion time when compared with the usual gas insufflation colonoscopy, especially considering that the www.giejournal.org

learning curve needs about 100 procedures to achieve sufficient skill, which may hamper its routine implementation.19,39,40 Our data show that WE cecal intubation time was significantly longer than that of the other techniques (ie, about 3 to 5 minutes), and yet withdrawal time was similar. When evaluating this additional time, it should be considered that in all selected studies, the WE technique was performed by adequately trained and motivated endoscopists; therefore, outside controlled studies, in daily practice, even longer procedure times might be expected, especially during the learning phase, and this could represent a critical issue. Indeed, WE might be less appealing in busy practices with high rates of deep sedation and with a tight schedule, where 3 to 5 additional minutes per colonoscopy would theoretically result in 1 fewer procedure every 6 or 8 procedures. Admittedly, the trade-off is between efficiency and institutional and social costs related to sedation, which could be lowered by decreasing recovery time and time until return to normal activity and by reducing use of sedation with cost savings freeing up most of the recovery area33; patient safety (presumably fewer sedation adverse events); fewer procedures rescheduled because of poor bowel preparation; but, most importantly, the increase in ADR (correlated to interval cancer and CRC deaths).16 It should be assessed whether the use of distal attachments for the colonoscope could increase ADR as WE does without lengthening total procedure time; however, the use of the device will increase cost. Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 7

WE colonoscopy increases adenoma detection rate

Fuccio et al

Our study has several strengths. Previous meta-analyses41-44 did not examine WI and WE separately or included a lower number of studies.45 An important addition to the current analysis is the inclusion of 3 recent RCTs comparing AI, WE,9,11,21 and WI9,11 with ADR as the primary outcome. We controlled that studies having the same recruitment period and same centers involved18,35 did not report overlapping data. We acknowledge the following limitations. First, the inherent indirectness of comparisons in the network meta-analysis may hamper the results: however, this should not be the case, because direct and network estimates were very close (Table 2 and Supplementary Table 1) and no inconsistency between direct- and indirect-only estimates was detected (Supplementary Fig. 4, available online at www.giejournal.org). Second, the limited number of studies that separately reported ADR according to indication hindered us to show significant differences in the right side of the colon ADR between WI and WE in the subgroup analyses and also suggests caution in interpreting our results. However, the low heterogeneity for primary outcomes indicates this should not be a major issue. Third, in most studies the endoscopists were not blinded to the outcomes, thus introducing potential bias. Fourth, direct comparisons regarding many secondary outcomes such as intubation and withdrawal time and real-time pain were affected by a high amount of heterogeneity, probably because of different endoscopists’ technical skills and case selection. Fifth, we did not stratify results according to CRC risk factors such as gender, because individual patient data were not reported by the included studies; however, this was not considered to be an issue because the proportion of male patients was indeed similar across most studies, ranging in most cases from 49% to 65% with only 4 studies reporting higher values. The implications for clinical practice are relevant. Until additional data are available, WE should be considered superior to WI if one is to forgo gas insufflation for colonoscope insertion and also wants to improve some key colonoscopy quality indicators. In addition, the primary outcome was the overall ADR, defined as ADR for procedures performed for any indication. This outcome should be more effective for establishing the mucosal inspection skills of colonoscopists and more representative of the procedures performed in everyday clinical practice.27 Some fields of research are left to investigate in the future, for example, the impact of WE on interval CRC and related deaths. It will be important to reproduce many key results about WAC at more centers. Future studies should consider recording polyps and/or adenomas removed on insertion, subtracting polypectomy time from total insertion time. Bowel cleanliness should be rated during insertion and during withdrawal, and the change in these scores would reflect the gain in colon preparation quality achieved by WE, which in turn could be correlated with an increase 8 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

in ADR or adenomas per colonoscopy. Finally, a different colonoscopist should perform the withdrawal; lack of blinding might bias bowel cleanliness score or where to locate a polyp, especially at the flexures. In conclusion, the current network meta-analysis shows that WE significantly increases the ADR in the entire and the right side of the colon, but with a significantly increased insertion time of about 3 to 5 minutes. Until additional data are available, WE should be considered superior to WI. Despite the large amount of available data, further studies coming from different groups of research are needed to confirm the current findings. Finally, the impact of the WE technique on interval cancer and related deaths should be ascertained.

REFERENCES 1. Cadoni S, Leung FW. Water-assisted colonoscopy. Curr Treat Options Gastroenterol 2017;15:135-54. 2. Radaelli F, Paggi S, Amato A, et al. Warm water infusion versus air insufflation for unsedated colonoscopy: a randomized, controlled trial. Gastrointest Endosc 2010;72:701-9. 3. Pohl J, Messer I, Behrens A, et al. Water infusion for cecal intubation increases patient tolerance, but does not improve intubation of unsedated colonoscopies. Clin Gastroenterol Hepatol 2011;9:1039-43. 4. Leung CW, Kaltenbach T, Soetikno R, et al. Water immersion versus standard colonoscopy insertion technique: randomized trial shows promise for minimal sedation. Endoscopy 2010;42:557-63. 5. Amato A, Radaelli F, Paggi S, et al. Carbon dioxide insufflation or warm-water infusion versus standard air insufflation for unsedated colonoscopy: a randomized controlled trial. Dis Colon Rectum 2013;56:511-8. 6. Falt P, Liberda M, Smajstrla V, et al. Combination of water immersion and carbon dioxide insufflation for minimal sedation colonoscopy. Eur J Gastroenterol Hepatol 2012;24:971-7. 7. Cadoni S, Sanna S, Gallittu P, et al. A randomized, controlled trial comparing real-time insertion pain during colonoscopy confirmed water exchange to be superior to water immersion in enhancing patient comfort. Gastrointest Endosc 2015;81:557-66. 8. Cadoni S, Falt P, Gallittu P, et al. Water exchange is the least painful colonoscope insertion technique and increases completion of unsedated colonoscopy. Clin Gastroenterol Hepatol 2015;13:1972-80. 9. Cadoni S, Falt P, Rondonotti E, et al. Water exchange for screening colonoscopy increases adenoma detection rate: a multicenter, doubleblinded, randomized controlled trial. Endoscopy 2017;49:456-67. 10. Hsieh Y-H, Koo M, Leung FW. A patient-blinded randomized, controlled trial comparing air insufflation, water immersion, and water exchange during minimally sedated colonoscopy. Am J Gastroenterol 2014;109: 1390-400. 11. Hsieh Y-H, Tseng C-W, Hu C-T, et al. Prospective multicenter randomized controlled trial comparing adenoma detection rate in colonoscopy using water exchange, water immersion, and air insufflation. Gastrointest Endosc 2017;86:192-201. 12. Leung FW. Water-aided colonoscopy. Gastroenterol Clin North Am 2013;42:507-19. 13. Wang X, Luo H, Xiang Y, et al. Left-colon water exchange preserves the benefits of whole colon water exchange at reduced cecal intubation time conferring significant advantage in diagnostic colonoscopy - a prospective, randomized controlled trial. Scand J Gastroenterol 2015;50:916-23. 14. Rex DK, Schoenfeld PS, Cohen J, et al. Quality indicators for colonoscopy. Gastrointest Endosc 2015;81:31-53.

www.giejournal.org

Fuccio et al 15. Corley DA, Jensen CD, Marks AR, et al. Adenoma detection rate and risk of colorectal cancer and death. N Engl J Med 2014;370:1298-306. 16. Kaminski MF, Wieszczy P, Rupinski M, et al. Increased rate of adenoma detection associates with reduced risk of colorectal cancer and death. Gastroenterology 2017;153:98-105. 17. Cadoni S, Gallittu P, Sanna S, et al. A two-center randomized controlled trial of water-aided colonoscopy versus air insufflation colonoscopy. Endoscopy 2014;46:212-8. 18. Leung FW, Harker JO, Jackson G, et al. A proof-of-principle, prospective, randomized, controlled trial demonstrating improved outcomes in scheduled unsedated colonoscopy by the water method. Gastrointest Endosc 2010;72:693-700. 19. Ramirez FC, Leung FW. A head-to-head comparison of the water vs. air method in patients undergoing screening colonoscopy. J Interv Gastroenterol 2011;1:130-5. 20. Cadoni S, Falt P, Sanna S, et al. Impact of colonoscopy insertion techniques on adenoma detection. Dig Dis Sci 2016;61:2068-75. 21. Jia H, Pan Y, Guo X, et al. Water exchange method significantly improves adenoma detection rate: a multicenter, randomized controlled trial. Am J Gastroenterol 2017;112:568-76. 22. Lee TJW, Rees CJ, Blanks RG, et al. Colonoscopic factors associated with adenoma detection in a national colorectal cancer screening program. Endoscopy 2014;46:203-11. 23. Harewood GC, Sharma VK, Garmo P de. Impact of colonoscopy preparation quality on detection of suspected colonic neoplasia. Gastrointest Endosc 2003;58:76-9. 24. Froehlich F, Wietlisbach V, Gonvers J-J, et al. Impact of colonic cleansing on quality and diagnostic yield of colonoscopy: the European Panel of Appropriateness of Gastrointestinal Endoscopy European multicenter study. Gastrointest Endosc 2005;61:378-84. 25. Lebwohl B, Kastrinos F, Glick M, et al. The impact of suboptimal bowel preparation on adenoma miss rates and the factors associated with early repeat colonoscopy. Gastrointest Endosc 2011;73:1207-14. 26. Moher D, Liberati A, Tetzlaff J, et al., and the PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Ann Intern Med 2009;151:264-9. 27. Rex DK, Ponugoti PL. Calculating the adenoma detection rate in screening colonoscopies only: is it necessary? Can it be gamed? Endoscopy 2017;49:1069-74. 28. Higgins JPT, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d5928. 29. Chen M-H, Shao Q-M. Monte Carlo Estimation of Bayesian credible and HPD intervals. J Comput Graph Stat 1999;8:69-92. 30. R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, 2017. ISBN 3-900051-07-0. Available at: http://www.R-project.org/. 31. Valkenhoef G van, Lu G, Brock B de, et al. Automating network metaanalysis. Res Synth Methods 2012;3:285-99. 32. Viechtbauer W. Conducting meta-analyses in R with the metafor package. J Stat Softw 2010;36:1-48. 33. Leung J, Mann S, Siao-Salera R, et al. A randomized, controlled trial to confirm the beneficial effects of the water method on U.S. veterans undergoing colonoscopy with the option of on-demand sedation. Gastrointest Endosc 2011;73:103-10.

www.giejournal.org

WE colonoscopy increases adenoma detection rate 34. Xu X, Zhu H, Chen D, et al. Carbon dioxide insufflation or warm-water infusion for unsedated colonoscopy: a randomized controlled trial in patients with chronic constipation in China. Saudi J Gastroenterol 2016;22:18-24. 35. Arai M, Okimoto K, Ishigami H, et al. A randomized controlled trial comparing water exchange and air insufflation during colonoscopy without sedation. Int J Colorectal Dis 2016;31:1217-23. 36. Garborg K, Kaminski MF, Lindenburger W, et al. Water exchange versus carbon dioxide insufflation in unsedated colonoscopy: a multicenter randomized controlled trial. Endoscopy 2015;47:192-9. 37. Kaminski MF, Thomas-Gibson S, Bugajski M, et al. Performance measures for lower gastrointestinal endoscopy: a European Society of Gastrointestinal Endoscopy (ESGE) Quality Improvement Initiative. Endoscopy 2017;49:378-97. 38. Yen AW. Insertion water exchange minimizes endoscopist multitasking during withdrawal inspection-a plausible explanation for enhanced polyp detection in the right colon. J Interv Gastroenterol 2015;5:3. 39. Wasan SK, Schroy PC. Water-assisted unsedated colonoscopy: Does the end justify the means? Gastrointest Endosc 2009;69:551-3. 40. Rex DK. Water exchange vs. water immersion during colonoscope insertion. Am J Gastroenterol 2014;109:1401-3. 41. Rabenstein T, Radaelli F, Zolk O. Warm water infusion colonoscopy: a review and meta-analysis. Endoscopy 2012;44:940-8. 42. Hu D, Xu Y, Sun Y, et al. Water infusion versus air insufflation for colonoscopy: a meta-analysis of randomized controlled trials. Tech Coloproctol 2013;17:487-96. 43. WU Jun, Bing HU. Comparative effectiveness of water infusion vs air insufflation in colonoscopy: a meta-analysis. Colorectal Dis 2013;15: 404-9. 44. Hafner S, Zolk K, Radaelli F, et al. Water infusion versus air insufflation for colonoscopy. Cochrane Database Syst Rev 2015;5:CD009863. 45. Leung FW, Amato A, Ell C, et al. Water-aided colonoscopy: a systematic review. Gastrointest Endosc 2012;76:657-66. 46. Caldwell DM, Gibb DM, Ades AE. Validity of indirect comparisons in meta-analysis. Lancet 2007;369:270; author reply 271. 47. Salanti G, Marinho V, Higgins JPT. A case study of multiple-treatments meta-analysis demonstrates that covariates should be considered. J Clin Epidemiol 2009;62:857-64.

Received January 24, 2018. Accepted June 21, 2018. Current affiliations: Department of Medical and Surgical Sciences, S. Orsola-Malpighi Hospital, University of Bologna, Bologna, Italy (1), Digestive Endoscopy Unit, Nuovo Regina Margherita Hospital, Rome, Italy (2), Department of Translational Research and New Technology in Medicine and Surgery, University of Pisa, Pisa, Italy (3), Digestive Endoscopy Unit, Division of Gastroenterology, Humanitas Research and University Hospital, Rozzano (MI), Italy (4), Division of Gastroenterology/Hepatology, Indiana University Hospital, Indianapolis, Indiana, USA (5), Digestive Endoscopy Unit, CTO Hospital, Iglesias, Italy (6). Reprint requests: Lorenzo Fuccio, MD, Gastroenterology Unit, DIMEC, University of Bologna, Via Massarenti 9, 40138, Bologna, Italy.

Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 9

WE colonoscopy increases adenoma detection rate

Fuccio et al

APPENDIX 1: METHODS Data sources and search strategy We performed a comprehensive literature search by consulting PubMed, EMBASE, SCOPUS, and the Cochrane Central Register of Clinical Trials (up to November 1, 2017) to identify full-text studies evaluating outcomes of various techniques of colonoscopic intubation (ie, water-aided vs air or CO2 insufflation). We also used ClinicalTrials.gov for ongoing or recently completed trials and PROSPERO for ongoing or recently completed systematic reviews. Electronic searches were supplemented by manual searches of references of included studies. The following medical subject headings (MeSH) and keywords were used to select studies: “water-aided colonoscopy,” “water exchange colonoscopy,” “water immersion colonoscopy,” “air insufflation colonoscopy,” “CO2 insufflation colonoscopy,” and “adenoma detection rate.” The search was circumscribed to English language. The Medline search strategy was (“Colon”[Mesh] OR “Colon”[All fields]) AND (“water”[All fields] OR “wateraided”[All fields]) AND “Adenoma”[All fields]) AND English[lang].

Selection process and data extraction Three independent reviewers assessed the eligibility of each study for inclusion, with any disagreements resolved by consensus assessment. The reasons for excluding trials were documented. The authors were not blinded to the journal titles or to the study authors or institutions.

9.e1 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

The following data were extracted for each study: publication status, publication date, study design and location, number of centers involved, study population, patient characteristics (average age, gender, previous abdominal surgery), ADR for the entire and proximal colon, overall and according to indication (ie, CRC screening vs diagnostic/surveillance setting), cecal intubation time, withdrawal time, cecal intubation rate, bowel cleansing according to BBPS both for the entire and for the right side of the colon, real-time insertion pain according to numerical scales, and proportion of unsedated patients.

Network meta-analysis The Markov Chain Monte Carlo approach was based on 4 chains and updated with 200,000 simulations, thinning 1 per 10 and a burn-in of 10,000. We produced an evidence network of colonoscopy techniques with results grouped for WI, WE, AI, and CO2 insufflation. We explored the comparison network of relative estimated effects by drawing the geometry of the 4 colonoscopy techniques, each presented as a node. The lines between any 2 nodes represent studies comparing 2 linked treatments. According to the transitivity assumption, the distribution of effect modifiers is considered to be the same in all pairwise comparisons. Violation of transitivity would be advised in case of inconsistency between direct and indirect results.46 Inconsistency is the difference between direct and indirect evidence for each linked comparison in the network, assessed by estimating concordance within closed loops.47 We further evaluated clinical and methodologic variables to identify possible causes of inconsistency. Publication bias was assessed with the network funnel plot.

www.giejournal.org

Fuccio et al

WE colonoscopy increases adenoma detection rate

PRISMA checklist for meta-analysis Section/topic

No.

Checklist item

Reported on page no.

1

Identify the report as a systematic review, meta-analysis, or both.

1

2

Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and implications of key findings; systematic review registration number.

3

Rationale

3

Describe the rationale for the review in the context of what is already known.

4-5

Objectives

4

Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, outcomes, and study design (PICOS).

5

Protocol and registration

5

Indicate if a review protocol exists, if and where it can be accessed (eg, web address), and, if available, provide registration information including registration number.

5

Eligibility criteria

6

Specify study characteristics (eg, PICOS, length of follow-up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale.

5-6

Information sources

7

Describe all information sources (eg, databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched.

Appendix

Search

8

Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated.

Appendix

Study selection

9

State the process for selecting studies (ie, screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis).

5-6

Data collection process

10

Describe method of data extraction from reports (eg, piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators.

Appendix

Data items

11

List and define all variables for which data were sought (eg, PICOS, funding sources) and any assumptions and simplifications made.

Appendix

Risk of bias in individual studies

12

Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.

7, Appendix

Summary measures

13

State the principal summary measures (eg, risk ratio, difference in means).

6-7

Synthesis of results

14

Describe the methods of handling data and combining results of studies, if done, including measures of consistency (eg, I2) for each meta-analysis.

6-7

Risk of bias across studies

15

Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies).

7, Appendix

Additional analyses

16

Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done, indicating which were pre-specified.

6-7, Appendix

Title Title Abstract Structured summary

Introduction

Methods

(continued on the next page)

www.giejournal.org

Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 9.e2

WE colonoscopy increases adenoma detection rate

Fuccio et al

Continued Section/topic

No.

Checklist item

Reported on page no.

Study selection

17

Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram.

7, Supplementary Figure 1

Study characteristics

18

For each study, present characteristics for which data were extracted (eg, study size, PICOS, follow-up period) and provide the citations.

Table 1

Risk of bias within studies

19

Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12).

Supplementary Table 3,4

Results of individual studies

20

For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group; (b) effect estimates and confidence intervals, ideally with a forest plot.

Table 1, Supplementary Tables 1 and 2

Synthesis of results

21

Present the main results of the review. If meta-analyses are done, include for each, confidence intervals and measures of consistency.

8-10

Risk of bias across studies

22

Present results of any assessment of risk of bias across studies (see Item 15).

10-11, Supplementary Tables 3 and 4

Additional analysis

23

Give results of additional analyses, if done (eg, sensitivity or subgroup analyses, meta-regression [see Item 16]).

Appendix 2

Summary of evidence

24

Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (eg, healthcare providers, users, and policy makers).

10

Limitations

25

Discuss limitations at study and outcome level (eg, risk of bias), and at review level (eg, incomplete retrieval of identified research, reporting bias).

13

Conclusions

26

Provide a general interpretation of the results in the context of other evidence, and implications for future research.

13

27

Describe sources of funding for the systematic review and other support (eg, supply of data); role of funders for the systematic review.

No funding

Results

Discussion

Funding Funding

From Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009;6:e1000097. For more information, visit www.prisma-statement.org.

9.e3 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

www.giejournal.org

Fuccio et al

WE colonoscopy increases adenoma detection rate

APPENDIX 2: RESULTS Risk of bias Overall, all studies were at a low risk of bias (Supplementary Table 3). After applying Cochrane’s collaboration tool for the primary outcome, there was no serious risk of bias for any of the direct comparisons. Heterogeneity analyses are illustrated in Supplementary Table 4. Overall, most comparisons yielded low heterogeneity. Inconsistency was not observed either for the primary outcome (Supplementary Fig. 4) or for the secondary outcomes. Thus, inconsistency was not considered to influence the effect size of the estimates. Regarding publication bias, visual inspection of our network funnel plot did not reveal marked asymmetry (Supplementary Fig. 5, available online at www. giejournal.org).

www.giejournal.org

Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 9.e4

WE colonoscopy increases adenoma detection rate

Fuccio et al

Identification

Literature search Period: through November 1st, 2017. Databases: PubMed, EMBASE, Scopus, Cochrane Central Register of Clinical Trials. Publication: Full text only. Language: English language only. Design: Randomized Controlled Trials.

Screening

155 studies were identified from databases

122 studies were excluded by titles

Eligibility

Irrelevant: 103 Reviews: 19

33 studies were retrieved for more details and application of inclusion criteria

16 studies were excluded ADR not reported: 8 Duplications: 8

Included

17 studies included 41 treatment arms 10,350 patients

Air Insufflation (AI) 15 studies included 15 treatment arms 3,975 patients

Carbon dioxide insufflation (CO2) 5 studies included 5 treatment arms 749 patients

Water Immersion (WI) 10 studies included 10 treatment arms 1,816 patients

Water Exchange (WE) 11 studies included 11 treatment arms 3,810 patients

Supplementary Figure 1. Study flow chart. ADR, Adenoma detection rate.

9.e5 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

www.giejournal.org

Fuccio et al

WE colonoscopy increases adenoma detection rate

CO2

WE

AI

WI Supplementary Figure 2. Network of comparisons for Bayesian metaanalysis. Each circle represents a colonoscopy technique. Each line represents direct comparison between two techniques. The width of the linking line is proportional to the number of studies. AI, Air; WI, water immersion; WE, water exchange.

Bayesian probability of ranking as the best technique 99%

98%

100% 90% 80%

75%

70% 60% 50%

50% 49%

46% 39%

40%

33%

31% 30%

30% 20%

16%

14% 5% 6%

10% 0%

1%

5%

1% 1%0%

1% 0% 0%

0%

0% Bowel cleansing, BBPS

Cecal intubation time

Withdrawal time

AI

CO2

Proportion of unsedated procedures

WI

Real-time pain, NRS/VAS

Adverse events rate

WE

Supplementary Figure 3. Bayesian probability of ranking as the best technique for secondary outcomes of the various colonoscopy techniques, as estimated by network meta-analysis. BPPS, Boston Bowel Preparation Scale; NRS/VAS, numeric rating scale/visual analog scale; AI, air; WI, water immersion; WE, water exchange.

www.giejournal.org

Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 9.e6

WE colonoscopy increases adenoma detection rate

Study P–value CO2 vs AIR direct indirect network

Fuccio et al

Odds Ratio (95% Crl)

.78525

0.98 (0.65-1.5) 1.1 (0.74-1.5) 0.95 (0.75-1.2)

.44325

1.4 (1.2-1.7) 1.1 (0.65-2.0) 1.4 (1.2-1.6)

WE vs AIR direct indirect network

WE vs CO2 direct indirect network

1.3 (0.97-1.8)

.367

1.7 (1.2-2.4) 1.5 (1.2-1.9)

WI vs CO2 direct indirect network

.1955

1.3 (0.94-1.9) 0.96 (0.66-1.4) 1.1 (0.88-1.5)

.86925

0.76 (0.60-0.94) 0.78 (0.58-1.1) 0.76 (0.64-0.90)

WI vs WE direct indirect network

0.5

1

3

Supplementary Figure 4. Inconsistency plot of the network split comparisons among the 4 colonoscopy techniques. Inconsistency p-values for each split comparison between direct and indirect comparisons are provided. P values > .05 denote the absence of inconsistency. WI, Water immersion; WE, water exchange; CrI, credible interval.

9.e7 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

www.giejournal.org

WE colonoscopy increases adenoma detection rate

0.242 0.485

0.363

Standard Error

0.121

0

Fuccio et al

-1

-0.5

0

0.5

1

Effect size centered at comparison-specific pooled efect Supplementary Figure 5. Network funnel plot for publication bias assessment. The dotted line implies the null hypothesis that the comparison-specific pooled effect estimates do not differ from the respective study-specific effect sizes.

www.giejournal.org

Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 9.e8

WE colonoscopy increases adenoma detection rate

Fuccio et al

SUPPLEMENTARY TABLE 1. Meta-analysis outcomes as obtained by direct pairwise meta-analysis Primary outcome (OR, 95% CI)

WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

ADR in entire colon, all settings

1.4 (1.2-1.6)

1.3 (.98-1.8)

1.32 (1.05-1.67)

1.0 (.87-1.2)

1.4 (.9-2.1)

.95 (.58-1.6)

ADR in right colon, all settings

1.57 (1.19-2.16)

d

1.4 (1.05-1.88)

1.13 (.84-1.53)

d

d

ADR in entire colon, CRC screening

1.50 (1.20-1.90)

d

1.37 (.91-2.17)

1.10 (.69-1.60)

d

d

ADR in entire colon, diagnostic/surveillance

1.60 (1.10-2.30)

d

1.25 (.63-2.50)

1.30 (.62-2.70)

d

d

ADR in entire colon, split-dose preparation

1.4 (1.2-1.7)

1.3 (.97-1.8)

1.33 (1.04-1.7)

1.1 (.9-1.4)

1.5 (.82-2.8)

1.3 (.48-3.5)

WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

Bowel cleansing, BBPS

.66 (.38-.97)

.01 (–.72 to .71)

.48 (.13-.81)

.32 (–.02 to .66)

d

d

Cecal intubation time, min

2.6 (.51-4.7)

1.8 (–2.5 to 6.0)

4.8 (1.8-7.8)

–.45 (–2.4 to 1.5)

.89 (–2.6 to 4.4)

–1.3 (–4.8 to 2.2)

–.06 (–.94 to .95)

.80 (–1.3 to 2.9)

–.30 (–2.2 to 1.6)

.78 (–.32 to 1.9)

.64 (–.76 to 2.2)

.10 (–1.4 to 1.6)

Secondary outcome (mean difference, 95% CI)

Withdrawal time, min Real-time insertion pain, NRS/VAS

–1.8 (–2.7 to –1.0) –.30 (–2.1 to 1.5) –.89 (–2.2 to .42) –1.7 (–2.5 to –.80) –.20 (–2.0 to 1.6) –2.8 (–4.7 to –.89) WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

Proportion of unsedated procedures

1.6 (.98-2.7)

.98 (.29-3.3)

1.49 (.4-3.33)

1.7 (.95-3.1)

1.2 (.3-4.8)

1.9 (.55-1.6)

Adverse events rate

1.1 (.35-3.6)

.2 (.1-4.5)

1.0 (.2-5.3)

.56 (.18-1.5)

.88 (.02-36.0)

.28 (.03-3.2)

Secondary outcome (OR, 95% CI)

Values are ORs or mean difference, along with 95% CI. Significant comparisons are in bold characters. ADR, Adenoma detection rate; BBPS, Boston Bowel Preparation Scale; CI, confidence interval; CRC, colorectal cancer; NRS/VAS, numeric rating scale/visual analog scale; OR, odds ratio; AI, air insufflation; WE, water exchange; WI, water immersion; d, not applicable.

9.e9 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

www.giejournal.org

Fuccio et al

WE colonoscopy increases adenoma detection rate

SUPPLEMENTARY TABLE 2. Raw estimates of meta-analysis outcomes, as computed by traditional meta-analysis Colonoscopy technique

Overall ADR in the entire colon (95% CI)

SUPPLEMENTARY TABLE 2. Continued Colonoscopy technique

Pooled mean real-time insertion pain (95% CI)

AI

4.6 (4.0-5.2)

31.1 (19.0-46.4)

CO2

4.1 (1.7-6.4)

WI

34.4 (28.3-40.9)

WE

2.8 (1.7-3.9)

WE

41.7 (32.5-51.5)

WI

AI

30.2 (24.4-36.8)

CO2

Colonoscopy technique AI

Overall ADR in the right side of colon (95% CI) 14.2 (11.6-17.3)

2.9 (2.3-3.5)

Colonoscopy technique

Proportion of unsedated procedures (95% CI)

AI

75.1 (45.5-91.6) 97.1 (82.8-99.6)

CO2

Lack of data

CO2

WI

15.8 (12.5-19.8)

WI

83.7 (63.9-93.7)

21.1 (16.4-26.8)

WE

86.7 (52.1-97.5)

WE Colonoscopy technique AI

ADR in the entire colon, CRC screening (95% CI)

Colonoscopy technique AI

Adverse events rate (95% CI) 1.3 (.5-3.3)

32.3 (23.7-42.3)

CO2

.7 (.2-2)

Lack of data

WI

1.6 (.7-3.3)

WI

39.5 (30-49.9)

WE

.6 (.2-1.6)

WE

42.4 (32.2-53.3)

CO2

Colonoscopy technique AI CO2

ADR all colon, surveillance/ diagnostic (95% CI) 20.5 (10.3-36.6) Lack of data

WI

44 (36.3-52)

WE

29.7 (15.7-49)

Colonoscopy technique AI

ADR in the entire colon, split-dose preparation (95% CI) 28 (20.6-36.8)

CO2

37.9 (21.3-58)

WI

37.6 (32.3-43.2)

WE Colonoscopy technique

40.7 (29.2-53.4) Pooled cecal intubation time, mean min (95% CI)

AI

10.2 (6.6-13.8)

CO2

8.5 (6.8-10.2)

WI

8.3 (7.4-9.2)

WE

13.4 (9.1-17.7)

Colonoscopy technique AI

Pooled withdrawal time, mean min (95% CI) 10.8 (9.3-12.2)

CO2

9.2 (7.7-10.8)

WI

10.8 (8.8-12.8)

WE Colonoscopy technique

11.5 (10.3-12.6) Pooled mean BBPS (95% CI)

AI

6.8 (6.3-7.3)

CO2

7.3 (7.0-7.6)

WI

7.0 (6.5-7.5)

WE

7.4 (7.0-7.8)

www.giejournal.org

Results are shown as proportions (categorical outcomes) or means (continuous outcomes), along with 95% confidence interval (CI). ADR, Adenoma detection rate; BBPS, Boston Bowel Preparation Scale; AI, air insufflation; WE, water exchange; WI, water immersion.

Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 9.e10

WE colonoscopy increases adenoma detection rate

Fuccio et al

SUPPLEMENTARY TABLE 3. Assessment of risk of bias Random sequence generation

Allocation concealment

Blinding of participants and personnel

Blinding of outcomes assessment

Incomplete outcome data

Selective reporting

Leung CW, 20104

þ

þ

-

-

þ

þ

Amato A, 20135

þ

þ

-

-

þ

þ

Radaelli F, 2010

þ

þ

-

-

þ

þ

Hsieh Y-H, 201410

þ

þ

-

-

þ

þ

Falt P, 20126

þ

þ

-

-

þ

þ

3

þ

þ

-

-

þ

þ

Xu X, 201634

þ

þ

-

-

þ

þ

Study

2

Pohl J, 2011

Ramirez J, 2011

-

-

-

-

þ

þ

Leung J, 201133

?

?

-

-

þ

þ

Arai M, 2016

þ

þ

-

-

þ

þ

Garborg K, 201536

þ

þ

-

-

þ

þ

Cadoni S, 201417

þ

?

-

-

þ

þ

19

35

Leung FW, 201018

?

?

-

-

þ

þ

Cadoni S, 20171

þ

þ

þ

þ

þ

þ

Hsieh YH, 2017

þ

þ

-

-

þ

þ

Jia H, 201721

þ

þ

-

-

þ

þ

?

?

-

-

þ

þ

11

20

Cadoni S, 2016

“þ” denote low risk of bias, “-“ indicate high risk of bias, and “?” represent unclear risk of bias.

SUPPLEMENTARY TABLE 4. Heterogeneity analysis for direct pairwise meta-analysis outcomes as expressed by Higgins’ I2 index Primary outcome (I2, %)

WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

ADR in entire colon, all settings

0%

0%

0%

30%

0%

0%

ADR in right colon, all settings

0%

d

0%

0%

d

d

ADR in entire colon, CRC screening setting

0%

d

14%

0%

d

d

ADR in entire colon, diagnostic/surveillance setting Secondary outcome (I2, %)

0%

d

0%

0%

d

d

WE vs AI

WE vs CO2

WE vs WI

WI vs AI

WI vs CO2

CO2 vs AI

Bowel cleansing, BBPS

89%

0%

18%

68%

d

d

Cecal intubation time, min

91%

97%

99%

94%

95%

94%

Withdrawal time, min

89%

0%

0%

40%

91%

57%

Real-time insertion pain, NRS/VAS

99%

0%

0%

73%

0%

0%

Proportion of unsedated procedures

73%

0%

0%

0%

0%

0%

Adverse events rate

0%

94%

0%

0%

0%

0%

An I2 >50% indicates high heterogeneity, whereas I2
50% denotes low heterogeneity. ADR, Adenoma detection rate; BBPS, Boston Bowel Preparation Scale; AI, air insufflation; WE, water exchange; WI, water immersion; NRS/VAS, numeric rating scale/visual analog scale; d, not applicable.

9.e11 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

www.giejournal.org