Endoscopy 2017; 49(08): 765-775
DOI: 10.1055/s-0043-105073
Original article
© Georg Thieme Verlag KG Stuttgart · New York

Adenoma detection with blue-water infusion colonoscopy: a randomized trial

Adriane Lesne1, 2, Olivier Rouquette3, Sandrine Touzet4, Fabien Petit-Laurent5, Gwenaelle Tourlonias6, Audrey Pasquion7, 8, Jérôme Rivory1, 6, Guillermo Aguero Garcete1, Julien Scanzi3, Sylvaine Chalumeau5, Christine Chambon-Augoyard1, Driffa Moussata9, Florence Leger-Nguyen5, Stéphane Degeorges10, Marion Chauvenet9, Thierry Fontanges10, Sandrine Baubet6, Philippe Brulet10, Claire Billioud6, Elsa Thimonier1, Karine Stroeymeyt-Martin9, Benjamin Hamel5, Emmanuelle Graillot1, Claire Cruiziat10, Olivia Scalone5, Marc O’Brien1, Denis Péré-Vergé8, Jean-Christophe Souquet6, Jean-Marc Phelip7, Laurent Poincloux3, Stéphanie Poupon-Bourdy4, Angélique Denis4, Laurent Magaud4, Thierry Ponchon1, 2, 11, Mathieu Pioche1, 2, 11
  • 1Hepatogastroenterology Department, Hôpital Edouard Herriot, Hospices Civils de Lyon, France
  • 2Lyon 1 University Claude Bernard, Lyon, France
  • 3Hepatogastroenterology Department, University Hospital d’Estaing, Clermont-Ferrand, France
  • 4Hospices Civils de Lyon, pôle Information médicale Evaluation Recherche, and Université de Lyon, EA 7425 Health Services and Performance Research (HESPER), Lyon, France
  • 5Hepatogastroenterology Department, Nord Ouest Hospital, Villefranche-sur-Saône, France
  • 6Hepatogastroenterology Department, University Hospital Croix Rousse, Lyon, France
  • 7Hepatogastroenterology Department, University Hospital Saint-Etienne, Saint-Priest en Jarez, France
  • 8Hepatogastroenterology Department, St Joseph St Luc Hospital, Lyon, France
  • 9Hepatogastroenterology Department, University Hospital Lyon Sud, Pierre-Bénite, France
  • 10Hepatogastroenterology Department, Public Hospital Bourgoin-Jallieu, France
  • 11INSERM U1032, LabTau, Lyon, France
Further Information

Corresponding author

Mathieu Pioche, MD PhD
Endoscopy unit - Digestive Disease Department
L Pavillon - Edouard Herriot Hospital
69437 Lyon
Fax: +33-4-72110147   

Publication History

submitted 05 May 2016

accepted after revision 15 February 2017

Publication Date:
11 April 2017 (eFirst)



Background and aims Colonoscopy is currently the reference method to detect colorectal neoplasia, yet some adenomas remain undetected. The water infusion technique and dying with indigo carmine has shown interesting results for reducing this miss rate. The aim of this study was to compare the adenoma detection rate (adenoma and adenocarcinoma; ADR) and the mean number of adenomas per patient (MAP) for blue-water infusion colonoscopy (BWIC) versus standard colonoscopy.

Methods We performed a multicenter, randomized controlled trial in eight units, including patients with a validated indication for colonoscopy (symptoms, familial or personal history, fecal occult blood test positive). Consenting patients were randomized 1:1 to BWIC or standard colonoscopy. All colonoscopies were performed by experienced colonoscopists. All colonoscopy quality indicators were prospectively recorded.

Results Among the 1065 patients included, colonoscopies were performed completely for 983 patients (514 men; mean age 59.1). The ADR was not significantly different between the groups; 40.4 % in the BWIC group versus 37.5 % in the standard colonoscopy group (odds ratio [OR] 1.13; 95 % confidence interval [CI] 0.87 – 1.48; P = 0.35). MAP was significantly greater in the BWIC group (0.79) than in the standard colonoscopy group (0.64; P = 0.005). For advanced adenomas, the results were 50 (10.2 %) and 36 (7.3 %), respectively (P = 0.10). The cecal intubation rate was not different but the time to cecal intubation was significantly longer in BWIC group (9.9 versus 6.2 minutes; P < 0.001).

Conclusion Despite the higher MAP with BWIC, the routine use of BWIC does not translate to a higher ADR. Whether increased detection ultimately results in a lower rate of interval carcinoma is not yet known.

Clinical trials registration: EudraCT 2012-A00548 – 35; NCT01937429.



Conventional colonoscopy with air insufflation (standard colonoscopy) is currently the reference method to detect and remove colorectal adenomas. It has demonstrated its efficacy to prevent colorectal cancer (CRC) and to reduce CRC-related mortality [1]. However, 0.3 % of patients undergoing colonoscopic surveillance will develop interval cancers related to missed adenomas [2]. On the other hand, tandem colonoscopy studies have reported an adenoma miss rate between 16 % and 24 % [3] [4] [5] [6] [7].

The adenoma detection rate (ADR; percentage of patients with at least one histologically confirmed colorectal adenoma or adenocarcinoma), as described by specialty societies [8] [9], is considered the prime quality indicator in CRC prevention because of its significant link with the risk of interval carcinomas [9] [10]. However, ADR does not measure the total number of adenomas detected within a person, resulting in a “one and done phenomenon” [11]; yet, if an appropriate surveillance interval is to be assigned after baseline colonoscopy, it would seem important that all polyps are identified during colonoscopy [12]. Although the risk of a diminutive adenoma developing into invasive carcinoma is small, the presence of three or more tubular adenomas predicts an increased risk of future advanced adenomas and CRC [13] [14] [15]. The European Society of Gastrointestinal Endoscopy (ESGE) guidelines therefore recommend a shorter interval of surveillance for such patients [16]. Consequently, the mean number of adenomas per patient (MAP) has been suggested in the literature to provide information additional to the ADR about colonoscopy performance [17].

Several factors have been considered as possible explanations for this adenoma miss rate, for example lesions behind colonic folds, colonic spasms, atypical flat lesions, short withdrawal time, and poor bowel preparation [18]. In the last 10 years, many endoscopic techniques [19] [20] [21] have been developed to improve the ADR, such as water infusion colonoscopy (WIC) and chromoendoscopy.

Chromoendoscopy, using indigo carmine, initially demonstrated controversial results with regard to ADR but recently a large randomized study observed a positive effect [22] [23]. Combining WIC and chromoendoscopy is possible by using water colored with indigo carmine and is called blue-water infusion colonoscopy (BWIC). Leung et al. [24] carried out a single-center study on American veterans in 2011 and reported an important benefit of BWIC on ADR. Larger studies in routine practice are needed to confirm this positive effect of BWIC versus standard colonoscopy on ADR and MAP in a non-academic setting that is closer to daily practice. In this context, we performed a multicenter randomized controlled study to compare ADR for BWIC versus standard colonoscopy.


Patients and methods


This study was performed between February 2013 and August 2014 in eight hepatogastroenterology units in France: five were in tertiary academic hospitals (university hospitals of Lyon Edouard Herriot, Lyon Croix Rousse, Lyon Sud, Saint-Etienne, and Clermont-Ferrand) and three were in local public (Villefranche and Bourgoin local hospitals) or private hospitals (Lyon Saint Luc Saint Joseph).


Trial design and approval

This study was a multicenter randomized study comparing standard colonoscopy and BWIC.

All documentation was submitted to and approved by the local ethics committee (“comité de protection des personnes” [CPP]) on February 25, 2014. All patients received an information letter and gave written consent in accordance with French ethics regulations. The National Advisory Committee on Information Processing in Material Research in the Field of Health also approved the study, with regard to the anonymous processing of personal health data. The study was approved by the medical ethics committee of all participating centers, no incentives were received by participating gastroenterologists, and the study underwent monitoring by an independent monitor. All authors had access to the study data and reviewed and approved the manuscript.


Trial registration

The GRAND BLEU trial was registered at the National French agency registration and European clinical trial register (EudraCT 2012-A00548-35) before the beginning of the study and then at the US National Institutes of Health (ClinicalTrials.gov: NCT01937429).



Consecutive patients scheduled for colonoscopy for one of the following indications were invited to participate: fecal occult blood test positive, personal or familial history of adenoma or cancer, hematochezia or digestive symptoms (in patients aged over 50), acromegaly, or endocarditis with digestive bacteria. Before inclusion, patient charts were reviewed for possible exclusion criteria: patients with polyposis syndromes, inflammatory bowel syndrome (IBD), previous partial or total colon resection, allergy to indigo carmine, or pregnancy. Patients were informed about the study during a consultation prior to their colonoscopy and received an information brochure to read at home. After they had given written informed consent, eligible patients were randomly allocated to either BWIC or standard colonoscopy.



Endoscopists with experience of more than 500 colonoscopies were enrolled to participate to this study. They received an extensive explanation and demonstration by the research coordinator but they had no clinical experience of BWIC prior to the beginning of the study. The colonoscopes used for the study were: CF-H180, CF-HQ190, PCF-H180 AL series with variable stiffness (Olympus Medical Systems, Tokyo, Japan) and Fujinon EC 530 WL instruments (Fujifilm Europe, Dusseldorf, Germany).

Patients received standard bowel preparation: low-residue diet and split oral intake of 4 L polyethylene glycol solution (Moviprep, Norgine, France or Fortrans, Ipsen, France). Procedures were performed with the patients spontaneously ventilating under sedation using intravenous propofol and remifentanil.

The intention of the endoscopists was to intubate the cecum directly without performing polypectomies during introduction. Cecal intubation was confirmed by documentation of landmarks (terminal ileum, appendiceal orifice, ileocecal valve). During withdrawal, regardless of the technique, the mucosa was carefully inspected using air insufflation and, if polyps were detected, endoscopic removal was attempted. If the cecum could not be reached for technical reasons, the procedure was considered a failure and the physician was free to use an air insufflation technique in the BWIC group or to use different endoscopes in either group.

All retrieved polyps were placed in separate containers and were sent for histopathological assessment. All polyp features (colonic segment, morphology [Paris classification], estimated size, optical diagnosis, and polypectomy technique) were reported. The intubation and withdrawal times (including time for polypectomies) were recorded by a member of the research staff, an endoscopy nurse, or by a gastroenterology fellow, using a stopwatch. Patient discomfort was scored with a numeric 0 to 10 scale by a member of research staff or a fellow after the procedure. Bowel preparation was scored per segment using the validated Boston Bowel Preparation score (BBPS) [25] ranging from 3 (complete segment without feces, good visualization) to 0 (solid feces, not evaluable).


Blue-water infusion colonoscopy (BWIC)

In WIC, air insufflation is replaced by infusion of water during colon intubation and the instilled water is removed during endoscope withdrawal. In this study, we chose to combine WIC with chromoendoscopy. A 0.008 % indigo carmine solution (10 mL 0.8 % indigo carmine diluted in 1 L warm water at a controlled temperature of 37 °C) was infused into the colon using a water pump controlled by a foot pedal. The air pump was turned off and only water infusion was used during insertion until cecal intubation had been achieved ([Fig. 1]; [Video 1]). Air insufflation was then turned on once the cecum had been reached. Water exchange, with suctioning of suspended residual feces and replacement by clean blue water, was performed during withdrawal. The volume of warm water infused was recorded.

Video 1 Example of the blue-water infusion colonoscopy procedure.

Georg Thieme Verlag. Please enable Java Script to watch the video.
Zoom Image
Fig. 1 Endoscopic images showing: a the appendix during blue-water infusion colonoscopy; b intubation using blue-water infusion colonoscopy; c,d adenomas that were detected by blue-water infusion colonoscopy.



Histopathology was processed and stained using standard methods and evaluated by one of the gastrointestinal pathologists in each center. The pathologists were blind to the study allocation. Polyp histology was evaluated according to the Vienna Criteria [26]. All lesions were classified as hyperplastic polyps; sessile serrated adenomas/polyps (SSAPs); traditional serrated adenomas; tubular, tubulovillous, or villous adenomas; or carcinoma. An advanced adenoma was defined as an adenoma ≥ 10 mm or with high grade dysplasia.



The primary outcome was adenoma detection, defined as the ADR i. e. the percentage of patients (per-patient analysis) with at least one histologically-confirmed colorectal adenoma or adenocarcinoma, as described by specialty societies [8] [9]. The MAP (per-polyp analysis) was also evaluated, defined by the total number of adenomas or adenocarcinomas detected in each group divided by the total number of patients in that group. 

Secondary outcomes included: cecal intubation rate and time, total procedure time, withdrawal time, resection time, size and location of adenomas or adenocarcinomas, and also patient discomfort and complications.

The SSAP detection rate was compared between the two groups and corresponded to the rate of patients with at least one SSAP.

In France, patients with at least one SSAP and/or at least three adenomas and/or the presence of high grade dysplasia or cancer and/or one adenoma > 10 mm are considered high risk patients. For these high risk patients, the interval before the follow-up colonoscopy should be 3 years and not 5 or 10 years as for lower risk patients. We also compared the rate of patients classified as high risk in the two groups.


Sample size determination

With a statistical power of 90 % and a two-sided alpha level of 0.05, the sample size needed to detect a 10 % absolute difference in ADR between the BWIC and standard colonoscopy groups was 476 per group, with an estimated 30 % ADR in the standard colonoscopy group [27] [28]. Taking into account, for each group, an intubation rate of 95 %, a 2 % rate of poor bowel preparation, and a 5 % rate of patients drop-out for various reasons (withdrawn consent, loss to follow-up, among others), we estimated the sample size needed to be 525 per group.



After having given written informed consent, eligible patients were randomly allocated 1:1 to either BWIC or standard colonoscopy by a computerized randomization program (Medsharing; Fontenay-sous-Bois, France). Randomization was stratified by center and random block sizes of 2, 4, 6, and 8 were used. Randomization was performed immediately after inclusion or within 24 hours prior to the colonoscopy by the research staff or the physician who enrolled the patient.


Adverse events

All procedural complications were recorded at the time of colonoscopy. Patients were instructed to contact the hospital if adverse events occurred later to ensure a complete adverse event record. The trial coordinator was informed of all adverse events, including delayed events, by written declaration.


Statistical methods

All data were analyzed using a modified intention-to-treat (ITT) approach. The modified ITT population comprised all randomized patients with endpoint data, i. e. who had undergone total colonoscopy allowing complete colon examination. Patients in whom cecal intubation failed were excluded from our analysis.

A sensitivity analysis was performed according to the strict ITT and per-protocol approaches. In the strict ITT approach, all randomized participants were analyzed, with zero being assigned as the number of detected adenomas for any randomized patients with missing adenoma data; these patients were excluded from the modified ITT. In the per-protocol analysis, all patients, even those with a protocol deviation (i. e. who received the other type of colonoscopy instead of their allocated colonoscopy), were included in the population. Given that the results from the ITT, modified ITT, and per-protocol populations were the same, the modified ITT results are presented in this manuscript, with the strict ITT results presented briefly in the sensitivity analysis section of the results.

Lesion detection rates were reported as frequencies and percentages, and numbers of lesions per patient as mean values and standard deviations. Hierarchical generalized linear models were used to estimate the differences between groups. Three-level random-intercept logistic (for lesion detection rate) and Poisson models (for number of lesions per patient) were fitted with a random intercept for endoscopists and a random intercept for centers to account for the nested nature of the data (polyps were nested in endoscopists, who were themselves nested in centers). Endoscopist and center were the clustering variables. The only predictor variable was the allocation group. 

The BWIC effect was given as the odds ratio (OR) for dichotomous outcomes or relative risk (RR) for count outcomes with a 95 % confidence interval (CI). Subgroup analysis was performed by the type of center (academic and non-academic) using hierarchical generalized linear models with adjustment for patient characteristics (age, sex, colonoscopy indication, colonoscopy withdrawal time, and BBPS score) and endoscopist experience. Models were fitted using the procedure Glimmix in the Statistical Analysis System (SAS) software (SAS Institute, Cary, North Carolina, USA).

The Mann – Whitney test was used to compare the BBPS score and procedure times between groups. The cecal intubation times in the BWIC group were compared between the first five colonoscopies performed by each endoscopist and their subsequent colonoscopies, using the Wilcoxon signed-rank test. Differences in size and location for small and diminutive adenomas or adenocarcinomas were compared between groups using a multivariable mixed model logistic model.

All analyses were performed using the SAS software version 9.3 (SAS Institute). A P value < 0.05 was regarded as significant. We did not modify the methods from the original declaration in the French national register of clinical trials, which was made before the first patient was included.



 [Fig.2] shows the study flowchart. Between February 2013 and August 2014, a total of 1065 patients consented to inclusion, although 15 were subsequently excluded. Of the 1050 randomized patients, 523 were allocated to standard colonoscopy and 527 to BWIC. As shown in  [Table 1], baseline characteristics were comparable according to age, sex, and colonoscopy indication.

Zoom Image
Fig. 2 Study flow chart. * Discontinued intervention for colonoscopy failure. Retaining allocation to groups according to randomization and including all subjects with endpoint information.
Table 1

Baseline characteristics of the patients who were randomized to undergo either blue-water infusion colonoscopy (BWIC) or standard colonoscopy.

BWIC group
(n = 492)

Standard colonoscopy group (n = 491)

Sex, male, n (%)

245 (49.8)

268 (54.6)

Median age (IQR), years

61.0 (52 – 67)

60.0 (51 – 66)

Colonoscopy indication, n (%)

  • Fecal occult blood test positive, n (%)

32 (6.5)

46 (9.4)

  • Family history of colorectal neoplasia, n (%)

188 (38.2)

170 (34.6)

  • Personal history of adenoma, n (%)

150 (30.5)

173 (35.2)

  • Digestive symptoms and aged ≥ 50, n (%)

135 (27.4)

126 (25.7)

  • Hematochezia and aged ≥ 50, n (%)

100 (20.3)

86 (17.5)

  • Acromegaly, n (%)

4 (0.8)


  • Endocarditis, n (%)

1 (0.2)

4 (0.8)

IQR, interquartile range.

Of the 1050 randomized patients, the primary outcome was not documented in 32 patients in the standard colonoscopy group and 35 in the BWIC group ([Fig. 2]). Finally, the primary outcome was documented in 983 patients (491 standard colonoscopy and 492 BWIC).

Colonoscopy results

The results of colonoscopy are presented in [Table  2]. Eighteen patients had an incomplete colonoscopy (10 standard colonoscopy and 8 BWIC). The cecal intubation rate was 98.1 % with BWIC and 98.4 % with standard colonoscopy (P = 0.83). The median time to cecal intubation was significantly longer with BWIC (9.9 minutes) than with standard colonoscopy (6.2 minutes; P < 0.001). Per operator, this median duration was not different for the first five BWIC procedures (12.8 minutes; interquartile range [IQR] 9.2 – 15.3) compared with the subsequent procedures (10.7 minutes; IQR 9.2 – 13.0; P = 0.29). The median withdrawal time was also significantly longer in the BWIC group (12.0 minutes) than in the standard colonoscopy group (11.0 minutes; P = 0.003). The mean volume of blue water infused during cecal intubation was 423.9 ± 230.1 mL. No significant differences were detected with respect to bowel preparation ([Table  2]).

Table 2

Comparison of results achieved with blue-water infusion colonoscopy (BWIC) and standard colonoscopy.

BWIC group
(n = 492)

Standard colonoscopy group
(n = 491)

P value

Median (IQR) Boston Bowel Preparation Scale (BBPS) score

8.0 (6 – 9)

8.0 (6 – 9)


Number of colonoscopies per center, n (%)

> 0.99

  • Center 1 – academic

147 (29.9)

141 (28.7)

  • Center 2 – academic

31 (6.3)

35 (7.13)

  • Center 3 – academic

65 (13.2)

63 (12.8)

  • Center 4 – academic

94 (19.1)

93 (18.9)

  • Center 5 – academic

17 (3.5)

20 (4.1)

  • Center 6 – non-academic

91 (18.5)

90 (18.3)

  • Center 7 – non-academic

26 (5.3)

28 (5.7)

  • Center 8 – non-academic

21 (4.3)

21 (4.3)

Cecal intubation rate, complete/total (%)

494 /502 (98.4)

492 /502 (98.0)


Insertion time, median (IQR), minutes

9.9 (6.8 – 14.6)

6.2 (4.5 – 9.3)

< 0.001

Withdrawal time, median (IQR), minutes

12.0 (8.0 – 20.1)

11.0 (7.2 – 16.0)


Examination time, median (IQR), minutes

10.1 (7.1 – 15.0)

9.1 (6.6 – 12.6)


Resection time, median (IQR), minutes

3.0 (1.0 – 7.0)

2.0 (1.0 – 4.0)


Total procedure time, median (IQR), minutes

24.6 (17.3 – 35)

19.0 (14.0 – 26.7)

< 0.001

IQR, interquartile range.


ADR and MAP outcomes

[Table 3 ] summarizes the results with regard to ADR and MAP. The ADR was not significantly different between the two groups: 40.4 % in the BWIC group versus 37.5 % in the standard colonoscopy group (OR 1.13; 95 %CI 0.87 – 1.48; P = 0.35). The MAP was significantly higher in the BWIC group at 0.79 ± 1.34 than in the standard colonoscopy group 0.64 ± 1.25 (P = 0.005).

Table 3

Comparison of lesions detected by blue-water infusion colonoscopy (BWIC) and standard colonoscopy.

BWIC group
(n = 492)

Standard colonoscopy group (n = 491)

OR/RR (95 %CI)

P value

Adenomas and/or adenocarcinomas

  • Detection rate (ADR), n (%)

199 (40.4)

184 (37.5)

1.13 (0.87 – 1.48)[1]


  • Number per patient (MAP), mean ± SD

0.79 ± 1.34

0.64 ± 1.25

1.24 (1.07 – 1.45)[2]



  • Detection rate, n (%)

196 (39.8)

184 (37.4)

1.11 (0.85 – 1.44)[1]


  • Number per patient, mean ± SD

0.78 ± 1.32

0.64 ± 1.25

1.23 (1.05 – 1.43)[2]


Advanced adenoma[4]

  • Detection rate, n (%)

50 (10.2)

36 (7.3)

1.48 (0.93 – 2.35)[1]


  • Number per patient, mean ± SD

0.13 ± 0.41

0.09 ± 0.33

1.49 (1.00 – 2.22)[2]


Three or more adenomas and/or SSAPs with dysplasia, and/or adenocarcinoma

  • Detection rate, n (%)

204 (41.5)

189 (38.5)

1.14 (0.8 – 7 – 1.48)[1]


  • Number per patient, mean ± SD

0.82 ± 1.37

0.66 ± 1.26

1.25 (1.08 – 1.45)[2]


Adenomas and/or SSAP and/or adenocarcinomas

  • Detection rate, n (%)

214 (43.5)

199 (40.5)

1.13 (0.87 – 1.48)[1]


  • Number per patient, mean ± SD

0.87 ± 1.44

0.70 ± 1.28

1.26 (1.09 – 1.46)[2]


Sessile serrated adenoma or polyp

  • Detection rate, n (%)

24 (4.9)

20 (4.1)

1.20 (0.64 – 2.22)[1]


  • Number per patient, mean ± SD

0.07 ± 0.35

0.05 ± 0.27

1.22 (0.73 – 2.06)[2]


Four or more lesions with dysplasia (high grade dysplasia and colorectal cancer)

  • Detection rate, n (%)

33 (6.7)

27 (5.5)

1.24 (0.73 – 2.12)[1]


  • Number per patient, mean ± SD

0.08 ± 0.31

0.06 ± 0.27

1.30 (0.80 – 2.10)[2]


OR, odds ratio; RR, relative risk; CI, confidence interval; ADR, adenoma detection rate; MAP, mean number of adenomas per patient; SD, standard deviation; SSAP, sessile serrated adenoma/polyp.

1 Three-level random-intercept logistic model with a random intercept for endoscopists and a random intercept for centers.

2 Three-level random-intercept Poisson model with a random intercept for endoscopists and a random intercept for centers.

3 Including low or high grade dysplasia but excluding adenocarcinoma.

4 Adenoma with high grade dysplasia or size ≥ 10 mm, but excluding adenocarcinoma (Vienne = 4 [without adenocarcinoma]).


Adenoma detection

Adenoma detection results are presented in [Table  3].

The proportion of patients with at least one adenoma (excluding adenocarcinoma) was not significantly different between the two groups: 39.8 % in the BWIC group versus 37.4 % in the standard colonoscopy group (OR 1.11; 95 %CI 0.85 – 1.44; P = 0.46). The proportion of patients with at least one advanced adenoma was similar between the groups: 10.2 % in the BWIC group versus 7.3 % in the standard colonoscopy group (RR 1.48; 95 %CI 0.93 – 2.35; P = 0.10). The mean number of advanced adenomas per patient was significantly different: 0.13 ± 0.41 in the BWIC group versus 0.09 ± 0.33 in the standard colonoscopy group (P = 0.048). Seven carcinomas were detected in the BWIC group versus two in the standard colonoscopy group (OR 3.52; 95 %CI 0.72 – 17.09; P = 0.12). For SSAPs, the mean number of lesions per patient was not significantly different: 0.07 ± 0.35 in the BWIC group versus 0.05 ± 0.27 in the standard colonoscopy group (P = 0.45). Using BWIC, the rate of high risk patients needing a shorter surveillance interval was significantly higher at 16.7 % versus only 10.8 % with standard colonoscopy (P = 0.008).

The sizes and locations of the detected adenomas or adenocarcinomas are presented in  [Table 4]. The size of adenomas or adenocarcinomas was not significantly different between the two groups; in particular, the detection of diminutive ( < 6 mm) adenomas or adenocarcinomas was not significantly better with BWIC (289 versus 215; P = 0.36). No significant difference was found between the groups with regard to the location of diminutive adenomas or adenocarcinomas (P = 0.15).

Table 4

Comparison of the sizes and locations for the small and diminutive adenomas or adenocarcinomas found by blue-water infusion colonoscopy (BWIC) and standard colonoscopy.

BWIC group

Standard colonoscopy group

P value[1]


Total number [2]



Size, n (%)

  •  < 6 mm

289 (74.3)

215 (68.3)


  • 6 – 9 mm

49 (12.6)

53 (16.8)


  •  ≥ 10 mm

46 (11.8)

32 (10.2)


Adenomas < 6 mm

Total number




  • Right colon, n (%)

137 (47.4)

87 (40.5)


  • Transverse colon, n (%)

61 (21.1)

49 (22.8)


  • Left colon and rectum, n (%)

91 (31.5)

79 (36.7)


1 Mixed effects logistic regression model according to the modified intention-to-treat approach.

2 20 adenomas with size information missing (15 in the BWIC group and 5 in the standard colonoscopy group).



The colonoscopies were performed by 53 senior endoscopists, of whom 30 (56.6 %) had less than 10 years of experience. In the BWIC group, 314 colonoscopies (63.8 %) were performed by physicians with less than 10 years of experience versus 317 colonoscopies (64.7 %) in the standard colonoscopy group (P = 0.78).

Of the 983 colonoscopies, 789 (80 %) were performed by just 23 endoscopists, who each contributed at least 17 colonoscopies to the study. The remaining 194 colonoscopies were performed by 30 endoscopists. Each endoscopist performed a similar mean number of BWIC (9.12 ± 10.9) and standard colonoscopy procedures (9.41 ± 10.2) within the study (P = 0.24). The study design did not allow the comparison of ADR among endoscopists. Looking at the physicians who performed more than 20 colonoscopies in each group (19/53; 35.8 %), the ADR per endoscopist ranged from 15.7 % to 81.0 % in the BWIC group and from 28.6 % to 49.0 % in the standard colonoscopy group.

The ADR in academic centers was greater than in the non-academic centers (42.2 % versus 30.7 %; P = 0.06). This difference was also non-significant after adjusting for patient age and sex, colonoscopy indication, colonoscopy withdrawal time, allocation group, and endoscopists’ experience (P = 0.22). The MAP was significantly higher in academic centers than in non-academic centers (0.84 ± 1.44 versus 0.40 ± 0.68; P = 0.04). Once patient and endoscopist characteristics had been controlled for, the MAP did not differ significantly between centers (P = 0.46).

The ADR (42.2 % versus 30.7 %; P < 0.001) and MAP (0.84 ± 1.44 versus 0.40 ± 0.68; P < 0.001) were significantly higher in academic centers than in non-academic centers; a similar result was found after adjusting for colonoscopy indication.

In colonoscopies performed in academic centers, the ADR did not differ between the BWIC (45.2 %) and standard colonoscopy (39.2 %) groups (P = 0.12), but the MAP was higher in the BWIC group (0.96 ± 1.48 versus 0.72 ± 1.40; P < 0.001). In procedures performed in non-academic centers, there were no differences in ADR or MAP between the BWIC and standard colonoscopy groups (28.3 % versus 33.1 %, respectively; P = 0.40; and 0.36 ± 0.66 versus 0.43 ± 0.70; P = 0.27).


Sensitivity analysis

By a strict ITT analysis, in which all randomized patients were analyzed, the ADR was not different between the two groups with 37.8 % in BWIC respectively versus 35.2 % in the standard colonoscopy group (OR 1.12; P = 0.41). Similarly, for advanced ADR, the results did not differ between the two groups with 9.5 % in the BWIC group versus 6.9 % in the standard colonoscopy group (P = 0.11). The MAP however was significantly increased in the BWIC group compared with the standard colonoscopy group at 0.74 versus 0.60 (P = 0.006).


Adverse events

The perforation rate was not significantly different between the two groups: one perforation with BWIC and one with standard colonoscopy (P > 0.99). No other adverse events related to the colonoscopy were reported in either of the two groups.



In this multicenter randomized controlled trial, we compared adenoma detection with BWIC versus standard colonoscopy and observed that BWIC led to a significant increase in the MAP (BWIC 0.79 ± 1.34 versus standard colonoscopy 0.64 ± 1.25; P = 0.005). More adenomas were detected but most of these were classified as non-advanced, and the number of patients with at least one detected adenoma (the ADR) was not significantly different (BWIC 40.4 % versus standard colonoscopy 37.5 %; P = 0.35).

The ADR is considered the most important surrogate measure for colonoscopy quality and was therefore chosen as the primary outcome in this study. Nevertheless, it has the inherent limitation that it does not measure the total number of adenomas detected during a procedure, which could provoke the “one and done” phenomenon [11]. Therefore, to avoid this, in addition to ADR, the MAP should also be described [17]. Furthermore, guidelines adapting the follow-up interval are based on the total number of adenomas and recommend a shorter interval for patients with more than three adenomas; a less intensive surveillance could result in interval carcinomas [11] [12], but the cost burden of such a strategy has not been fully determined. It is therefore of note that, in this study, high risk patients needing 3 years between colonoscopies were detected significantly more frequently with BWIC than with standard colonoscopy (16.7 % versus 10.8 %, respectively; P = 0.008).

This is one of the first large prospective randomized controlled trials that has been adequately powered to compare adenoma detection between BWIC and standard colonoscopy. Colonoscopies were performed by experienced endoscopists in five academic centers and three regional hospitals. The BWIC and standard colonoscopy techniques were consecutively performed in a random order and data on polyp detection, procedure times, and bowel preparation scores were prospectively recorded, ensuring accurate and optimal data collection of colonoscopies. We therefore believe that our results are reliable and applicable to daily clinical practice.

Some limitations of the study have to be acknowledged. Our sample size calculation was based on an ADR of 30 % in the standard colonoscopy group; however, the ADR in the standard colonoscopy group was 37 %, which is much higher than expected as a previous meta-analysis had a lower baseline ADR of 31 % for standard colonoscopy [29]. Because this was the first large prospective colonoscopy study performed by our research group, we could not have anticipated such a high ADR in the standard colonoscopy group.

Despite the high ADR in the standard colonoscopy group, the ADR in the BWIC group was 40.4 %, a relative increase of 7.9 % compared with that found in the standard colonoscopy group, although this difference was not significant. The advanced adenoma ADR was also higher at 10.2 % in the BWIC group, which is a relative increase of 38.6 % compared with 7.3 % in the standard colonoscopy group. The main limitation, as mentioned before, was the lack of previous training of the operators in the BWIC technique, which could be, at least in part, responsible for the increased intubation time. This lack of experience with BWIC could also have introduced a learning curve in the detection rate. Nevertheless, BWIC did not materially change the detection technique for experienced endoscopists because the withdrawal technique was close to that found for standard colonoscopy and the blue coloration was very light. It would have also been interesting to include a third group of patients who would have received WIC without indigo carmine to evaluate the clear impact of adding a small amount of this dye for detection, but the number of patients needed would have been too high for this to have been feasible in the participating centers.

In our study, operators could not be blinded to the allocated technique. The ADR varied from 15.7 % to 81.0 % with large differences between endoscopists, which is also a limitation, but has been described in several previous studies [10] [11] [30]. Nevertheless, some endoscopists had very low detection rates compared with the 38.6 % mean ADR for the study, thereby introducing a dispersion bias. Another limitation was the heterogeneous number of colonoscopies performed by each physician who participated in the study as only 35.8 % (19/53) of the endoscopists performed more than 20 procedures in the study.

The 24 mistakes in allocation to the correct study arm are another limitation, but these mistakes were made for organizational reasons (poor transmission of information about the allocated arm) and were not induced by patient characteristics. This limitation did not modify the results in the per-protocol analysis.

In this study, we found high ADRs of 37.5 % and 40.4 % in the standard colonoscopy and BWIC groups, respectively. Initially, water-aided methods, including WIC and water exchange colonoscopy (WEC), were evaluated to minimize abdominal pain and discomfort during unsedated colonoscopy [31] [32] [33]. It later appeared that the WIC technique could also enhance the ADR [34] [35] [36]. Several large randomized controlled trials and a meta-analysis have compared the ADR determined by WIC and by standard colonoscopy. Although the earliest meta-analysis showed a slightly increased ADR using WIC with respect to standard colonoscopy (RR 1.16) [29], this higher level of detection could be partly explained by the significantly better tolerance of WIC in unsedated patients.

With regard to BWIC versus standard WIC, a randomized controlled trial demonstrated a significant difference using blue water, despite a relatively small number of included patients (n = 168); the use of water colored by minute amounts of indigo carmine dye increased the ADR from 44 % to 62 % (P = 0.03) [36]. The concentration of indigo carmine (0.008 %) used in the BWIC technique was dramatically lower than that used in typical chromoendoscopy with indigo carmine, as recommended in patients with IBD or Lynch syndrome [37] [38]. In screening patients, indigo carmine chromoendoscopy did not demonstrate increased adenoma detection compared with standard colonoscopy [39]. To the best of our knowledge, our study is the first randomized controlled trial comparing BWIC and standard colonoscopy.

As previously demonstrated in a meta-analysis, the cecal intubation rate is not different between standard colonoscopy and WIC, but the cecal intubation time is significantly longer (+ 1.09 minutes) in WIC with a high level of heterogeneity between studies [29]. In the present trial, without training in BWIC being given before starting the study, the cecal intubation time in this group was significantly increased compared with standard colonoscopy (median 9.9 versus 6.2 minutes) but the cecal intubation rate was not significantly different.

The median duration of withdrawal for BWIC was also increased (12.0 versus 11.0 minutes). This difference can partly be attributed to the increased number of resected adenomas in the BWIC group and the need to suction the water instilled during intubation. This longer withdrawal time could introduce a bias because detection is linked to withdrawal time, as has been demonstrated in previous studies [8] [40]. Nevertheless, we adjusted our comparison model by introducing the withdrawal time to reduce this bias. Unfortunately, the previous meta-analysis comparing WIC and standard colonoscopy did not analyze withdrawal times, which precluded comparison.

The number of small and large adenomas did not significantly increase with BWIC, nor did the number of advanced adenomas and right-sided diminutive adenomas. Previous studies on WIC and BWIC did not separately analyze diminutive polyps [24] [29] [36] [41]. As BWIC seems only to result in the detection of more adenomas (resulting in an increased MAP) and not in greater detection of advanced adenomas, its clinical value in routine practice may be questioned. Although no data are available as to the association between MAP and interval carcinomas, we anticipate that such a link could be detected in future studies as a higher MAP is linked to a more precise examination of the colon; future large epidemiological studies are needed to investigate this.

Whether the benefits of increased detection of adenomas in patients, without a significant increase in ADR or detection of advanced adenomas, justify changing the current colonoscopy technique to BWIC may vary between centers and settings. The routine or selective use of BWIC will be guided by budget considerations as the use of indigo carmine and the longer procedure time increase the cost of the procedure. The relative weight one attaches will also guide this use of BWIC to improve detection as an indicator of colonoscopy quality and to improve the long-term protection of patients against CRC.


Competing interests



This study was funded by a grant from the French Ministry of Health (Programme Hospitalier de Recherche Clinique Inter-régional 2011: IR-27 – 20). It also received the “Association pour la recherche sur le cancer” prize, the “Ligue contre le cancer” prize, and the French national gastroenterology society research grant in 2014 (Société française de gastroentérologie: AO FARE 2014).

We sincerely thank all the physicians who participated in the study and the clinical research team from the Hospices Civils de Lyon who performed the data collection, control, and analysis. We particularly thank Philip Robinson who reviewed this paper. We thank Albert Mvondo and Sophie Hommey for their help with data collection and all the residents who collaborated on this project.

We also acknowledge Life Partners Europe for providing the indigo carmine blue for the study.

Corresponding author

Mathieu Pioche, MD PhD
Endoscopy unit - Digestive Disease Department
L Pavillon - Edouard Herriot Hospital
69437 Lyon
Fax: +33-4-72110147   

Zoom Image
Fig. 1 Endoscopic images showing: a the appendix during blue-water infusion colonoscopy; b intubation using blue-water infusion colonoscopy; c,d adenomas that were detected by blue-water infusion colonoscopy.
Zoom Image
Fig. 2 Study flow chart. * Discontinued intervention for colonoscopy failure. Retaining allocation to groups according to randomization and including all subjects with endpoint information.