Real-world evidence of water exchange colonoscopy improving adenoma detection rates: A propensity score-matched analysis Authors: Ying-Ju Chao, Yu-Hsi Hsieh, Chia-Pei Tang, Ping-Hung Ko, Yen-Chun Chen, Chia-Hsin Cheng, Chih-Wei Tseng Published: 2026 ABSTRACT BackgroundRandomized trials show that water exchange (WE) colonoscopy improves adenoma detection rate (ADR), but its effectiveness and workflow trade-offs in routine, high-volume practice remain uncertain.ObjectivesTo evaluate the real-world performance of WE versus conventional air insufflation for adenoma detection, bowel preparation quality, and procedure efficiency.DesignRetrospective propensity score matched cohort study.MethodsOutpatient colonoscopies performed from January 2019 through October 2024 at a high-volume Taiwanese center were analyzed. WE procedures were matched 1:1 with air insufflation controls on prespecified covariates. The primary outcome was ADR. Secondary outcomes included segment-specific detection, advanced adenomas detection, sessile serrated lesion detection, bowel preparation quality, and procedure times.ResultsThe matched cohort included 4,322 colonoscopies (2,161 per group). WE increased ADR by 14.0 percentage points (57.9% vs 44.0%; P < 0.001), corresponding to a number needed to treat of 8. The largest gain occurred in the right colon (right-sided ADR 31.8% vs 18.0%; P < 0.001). WE increased detection of adenomas ≥ 10 mm (8.7% vs 6.7%; P = 0.014) and sessile serrated lesions (1.9% vs 1.0%; P = 0.011). WE improved bowel preparation (adequate preparation 97.4% vs 88.4%; P < 0.001; total BBPS 7.5 vs 6.4; P<.001). Total procedure time was longer with WE (22.6 vs 15.6 minutes; P < 0.001), primarily due to longer insertion (11.0 vs 6.0 minutes; P < 0.001).ConclusionsIn real-world, high-volume practice, WE substantially improved neoplasia detection and bowel cleansing, with the greatest benefit in the right colon and persistence of effect against a high-performing air insufflation baseline. The time cost was predictable, supporting WE as an effective insertion technique when maximizing diagnostic yield is prioritized. Introduction Colorectal cancer (CRC) is the third most common cancer worldwide, with an estimated 2.29 million new cases, and the second leading cause of cancer-related death, accounting for 1.11 million deaths annually. 1 CRC mortality is projected to increase by 93.5% to 2.21 million deaths by 2050. 1 In the face of this rapidly rising burden, effective screening and prevention strategies are increasingly important. 2 Most sporadic CRCs arise from identifiable precursor lesions, approximately 70% from conventional adenomatous polyps and about 25% from sessile serrated lesions (SSLs). 3 Notably, SSLs are more frequently located in the right colon and are often flat, subtle, and obscured by residual stool or mucus, making the right colon a critical site for missed lesions and post-colonoscopy colorectal cancer (PCCRC).3–5 Accordingly, colonoscopy, by enabling the detection and removal of precancerous lesions, is widely recommended to reduce CRC incidence and mortality. 6 The adenoma detection rate (ADR) is a cornerstone quality indicator of colonoscopy and is inversely associated with CRC risk; it also correlates with prevalent CRC detection.7,8 Despite ongoing quality-improvement efforts, a recent systematic review using the WEO methodology reported that the pooled 3-year PCCRC rate in Western countries remained as high as 7.5% (95% CI, 6.4%–8.7%). 9 Therefore, improving ADR remains a critical objective for all colonoscopists, especially for those with baseline ADRs below 26%, who appear to gain the greatest reduction in CRC incidence from incremental improvements in detection performance. 10 Among strategies developed to improve ADR, water-aided colonoscopy, particularly water exchange (WE), has received increasing attention because it may enhance mucosal visualization.11,12 Compared with conventional air insufflation, WE is an alternative insertion technique designed to improve mucosal cleanliness and enhance lesion detection by using continuous water infusion with near-complete suction removal of luminal contents during insertion. 13 This mechanism may be particularly relevant in the right colon, where residual debris, mucus, and subtle serrated lesions can impair detection.14,15 Randomized trials and meta-analyses have shown that WE increases overall ADR, advanced adenoma detection rate (advanced ADR), and serrated lesion detection compared with air insufflation.13,16–19 WE has also been associated with lower right-colon adenoma and serrated lesion miss rates compared with air insufflation or carbon dioxide insufflation techniques.14,15 Many studies were conducted under highly standardized conditions that may not reflect everyday clinical settings. Specifically, prior evidence is limited by the exclusion of patients at the extremes of age or with complex comorbidities, and a lack of data from high-volume centers with routine workflow constraints.17,20,21 Furthermore, whether WE maintains its incremental benefit for right-sided ADR and sessile serrated lesion detection against a high-performing baseline remains unclear.14,22 Prior trials also avoided the selection bias inherent in non-randomized practice, such as patient-driven self-payment. Therefore, real-world data are needed to clarify the effectiveness of WE in clinical practice. Therefore, we conducted this real-world observational study to test the predefined hypothesis that water exchange is associated with a higher ADR than air insufflation, and to evaluate whether this association is consistent across secondary detection outcomes and clinically relevant subgroups. Because the choice of insertion method in routine practice is not randomized and may be influenced by baseline confounding factors, we used propensity score matching (PSM) to balance measured baseline characteristics between groups and to mitigate confounding by indication. Specifically, the propensity model incorporated key clinical and procedural covariates known to influence ADR based on prior literature, including patient age, sex, body mass index (BMI), baseline bowel preparation regimen, FIT/FOBT positivity, prior abdominal surgery, constipation, smoking, and drinking. 23 Materials and methods Patients and study design This retrospective propensity score–matched study was conducted at Dalin Tzu Chi Hospital, Taiwan, and included colonoscopies performed between January 2019 and October 2024 (Figure 1). The reporting of this observational cohort study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for cohort studies. 24 A completed STROBE checklist is provided as Supplementary Table S1, and the study flow diagram is presented in Figure 1. We included outpatient colonoscopies performed for screening, surveillance, a positive fecal immunochemical test (FIT)/fecal occult blood test (FOBT), or diagnostic indications in patients aged 16–102 years. Among 6,305 colonoscopies identified, we excluded procedures in patients with prior colectomy (n=48), those performed in inpatient or emergency department settings (n=448), procedures in patients with known CRC at the time of colonoscopy (n=232), and procedures performed using the water-immersion technique (n=225). The remaining 5,352 colonoscopies constituted the analytic cohort (Figure 1).Figure 1.Cohort selection and propensity score matching. Cohort selection and propensity score matching. Exposure classification and propensity score matching In routine practice, the choice of insertion method was not randomized. Water exchange (WE) was available as an optional self-pay insertion technique (US$15), and allocation to WE or conventional air insufflation was determined by patient preference after counseling. Eligible procedures were classified according to the insertion method into the air insufflation group (n=2,169) and the WE group (n=3,183) (Figure 1). To mitigate confounding due to non-random selection, we performed 1:1 PSM between the WE and air insufflation groups. After matching, colonoscopies remained in each group for outcome comparisons (air insufflation matched n=2,161; WE matched n=2,161). Outcome variables and definitions The primary outcome was adenoma detection rate (ADR), defined as the proportion of colonoscopies in which at least one histologically confirmed conventional adenoma was detected. 25 Formal mucosal inspection was performed during withdrawal after cecal intubation, and outcome measurement was based on lesions detected and removed during the procedure. To capture all eligible real-world data, the unit of analysis was the colonoscopy procedure rather than the individual patient; therefore, repeat procedures in the same patient were eligible for inclusion if they independently met the predefined selection criteria. Secondary outcomes included polyp detection rate (PDR), SSL detection rate, advanced ADR, adenomas per colonoscopy (APC), and adenomas per positive colonoscopy (APP). Segment-specific outcomes were also evaluated, including right-sided, transverse, and left-sided ADR, advanced ADR, and SSL detection rate. Advanced adenoma was defined as villous histology, high-grade dysplasia, or size ≥10 mm. Pathologic assessments were performed by pathologists blinded to the insertion method. In contrast, endoscopists were not blinded to the insertion method because of the retrospective real-world study design. Baseline variables extracted from electronic medical records included sex, age, BMI, smoking, alcohol use, FIT/FOBT positivity, constipation, prior abdominal surgery, sedation status, and bowel preparation regimen; these variables were used for PSM. The elderly population was defined as individuals aged 65 years and older. 26 BMI was calculated using the formula: BMI = body weight (kg)/[height (m)]. 2 According to Taiwan’s recommendation, a BMI of 24 or higher is classified as overweight. 27 Smoking and alcohol use were recorded as ever use based on structured preprocedural assessment. Sedation was offered based on patient preference and clinical assessment. Sedated procedures utilized a standardized propofol-based intravenous protocol across both groups, consisting of an initial bolus of 1 mg/kg, or 0.5 mg/kg for patients older than 65 years, followed by 20 to 30 mg titrations 28 Sedation status was recorded as a binary variable and included in both the propensity score model and the multivariable logistic regression analysis. Procedure performance metrics included cecal intubation rate, cecal insertion time, withdrawal time, and total procedure time (including polypectomy time). Bowel preparation quality was assessed using the Boston Bowel Preparation Scale (BBPS), reported as total and segmental scores (0–3 per segment). Adequate bowel preparation was defined as a total BBPS score ≥6 with each segment score ≥2. 25 An adequate withdrawal time for colonoscopy was defined as a minimum of 9 minutes.29,30 Data were extracted from structured colonoscopy reports and electronic medical records, where completion of the report required entry of predefined mandatory fields. Consequently, no missing data were observed for the analyzed variables, and all records were included without imputation. Preparation and procedure for colonoscopy All procedures were performed by three experienced endoscopists (YHH, CWT, and CPT), each with more than 15 years of experience and having performed over 1,000 WE colonoscopies. In alignment with international quality standards, the operators routinely targeted a minimum withdrawal time of 6 minutes for negative screening examinations. Standardized inspection techniques, including meticulous fold examination and fluid suctioning, were universally expected across all procedures. Bowel preparation was administered using either split-dose or single-dose regimens, including a 2-L polyethylene glycol (PEG) solution combined with 10 mg bisacodyl, or sodium picosulfate/magnesium oxide/citric acid (Boklean®) combined with 10 mg bisacodyl. Examinations were initiated with the patient in the left lateral decubitus position. Standard loop-reduction maneuvers were applied when scope advancement was impeded or paradoxical movement occurred. Cecal intubation was confirmed by identification of the appendiceal orifice and/or the ileocecal valve; successful intubation was defined as the ability to touch the cecal base with the colonoscope tip. Adult video colonoscopes without variable stiffness (CF-H260 and CF-HQ290; Olympus, Tokyo, Japan) were used for all procedures. In the air-insufflation group, air was used as needed to minimally distend the colonic lumen during insertion. In the WE group, the technique followed a standardized institutional protocol based on international consensus recommendations. 13 The air pump was turned off and no air insufflation was used during insertion until cecal intubation was achieved. Warm tap water was infused through the accessory channel using a foot-switch–controlled water pump (OFP-2; Olympus, Tokyo, Japan). During insertion, infused water was predominantly removed concurrently by suction, with aspiration of suspended fecal material and residual gas pockets or bubbles, followed by infusion of clean water to improve mucosal visualization and facilitate scope advancement. The amount of water infused was adjusted according to luminal cleanliness and visibility rather than a fixed volume. However, the procedural goal was near-complete water exchange, with infused and aspirated water volumes approximately balanced by the time of cecal intubation. This uniform protocol was strictly adhered to by all participating endoscopists. During withdrawal, air insufflation was used in both groups to achieve adequate luminal distension for careful mucosal inspection. Sample size calculation Because of the retrospective study design, no formal a priori sample-size calculation was performed. Instead, all eligible consecutive outpatient colonoscopy procedures performed during the prespecified study period were included to maximize statistical precision and minimize selection bias. Statistical analysis PSM was used to reduce confounding due to non-random selection of insertion method. Propensity scores were estimated using logistic regression with insertion method (WE vs air insufflation) as the dependent variable and prespecified covariates as predictors, including age, sex, BMI, sedation status, bowel preparation regimen, smoking, alcohol use, FIT/FOBT positivity, prior abdominal surgery, and constipation. Covariates were selected a priori based on clinical relevance and prior evidence of their associations with ADR. 23 Only variables available before or at the time of colonoscopy were included. Therefore, the baseline bowel preparation regimen, rather than the final BBPS, was used in the propensity score model because BBPS was assessed during colonoscopy and may partly reflect the intra-procedural cleansing effect of WE. Two groups were matched 1:1 using nearest-neighbor matching without replacement, with a caliper width of 0.01 of the standard deviation of the logit of the propensity score. Covariate balance before and after matching was assessed using standardized mean differences (SMDs), with SMD <0.10 indicating adequate balance. All analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, USA). Categorical variables are presented as counts and percentages, and continuous variables as mean ± standard deviation (SD). In the unmatched cohort, categorical variables were compared using the chi-square test, and continuous variables using the Mann–Whitney U test. In the matched cohort, paired comparisons were performed using McNemar’s test for categorical variables and the Wilcoxon signed-rank test for continuous variables. For matched analyses, absolute effect sizes were reported as risk differences (RDs) for binary outcomes and mean differences (MDs) for continuous outcomes, each with 95% confidence interval (CIs). For clinically interpretable binary outcomes, the number needed to treat (NNT) was calculated as 100 divided by the absolute RD. NNT was not calculated for continuous outcomes or for binary outcomes with 95% CIs crossing the null. To identify independent predictors of ADR, univariate and multivariate logistic regression with forward stepwise selection was used. Insertion method, gender, and age were forced into the model a priori. Other candidate variables with a univariable P < 0.05 entered the stepwise pool (entry: P < 0.05, removal: P > 0.10). All tests were two-sided, and P <0.05 was considered statistically significant. ADR was the prespecified primary outcome. Secondary, segmental, and component-specific outcomes were considered exploratory. No formal adjustment for multiple comparisons was performed; therefore, P values for these analyses should be interpreted as nominal. Results Baseline characteristics A total of 6,305 colonoscopies were performed between January 2019 and October 2024. After the exclusion of 953 procedures failing to meet the inclusion criteria, 5,352 procedures were included in the analytic cohort. PSM (1:1) was performed using age, sex, BMI, sedation status, bowel preparation regimen, smoking, alcohol use, FIT/FOBT positivity, prior abdominal surgery, and constipation. The propensity score–matched cohort comprised 4,322 procedures, evenly distributed between the air insufflation group (n=2,161) and the WE group (n=2,161) (Figure 1). Table 1 summarizes baseline characteristics in the original and matched cohorts. In the original cohort, the groups differed in sedation status, bowel preparation regimen, alcohol use, and constipation. After matching, baseline characteristics were well balanced between groups, with all SMD below 0.1.Table 1.Baseline characteristics of the air insufflation and water exchange groups in the original and matched cohorts.Baseline characteristicsOriginal cohort (n=5,352)Matched cohort (n=4,322)Air insufflation (n =2,169)Water exchange (n =3,183)P Value|SMD| # Air insufflation (n =2,161)Water exchange (n =2,161)P Value|SMD| # Male, n (%)1,088 (50.2)1,620 (50.9)0.598 b 0.0151,085 (50.2)1,084 (50.2)1.000 c 0Age (years), mean (SD)59.0 (13.6)59.3 (12.3)0.910 a 0.01459.1 (13.6)59.3 (12.3)0.770 d 0.012BMI (kg/m2), mean (SD)25.3 (3.9)25.1 (3.7)0.240 a 0.03525.2 (3.9)25.2 (3.7)0.519 d 0.016Fully sedated colonoscopy, n (%) @ 1,379 (63.6)1,713 (53.8)<0.001 b 0.2241,371 (63.4)1,337 (61.9)0.107 c 0.035Bowel preparation with PEG, n (%)1,111 (51.2)1,796 (56.4)<0.001 b 0.1151,111 (51.4)1,110 (51.4)1.000 c 0Smoking, n (%)*218 (10.1)369 (11.6)0.076 b 0.086218 (10.1)230 (10.6)0.566 c 0.030Drinking, n (%)*202 (9.3)364 (11.4)0.013 b 0.125202 (9.3)210 (9.7)0.678 c 0.025FIT/FOBT positive, n (%)307 (14.2)510 (16.0)0.062 b 0.078307 (14.2)318 (14.7)0.658 c 0.022Previous abdominal surgery, n (%)330 (15.2)520 (16.3)0.270 b 0.046330 (15.3)342 (15.8)0.643 c 0.021Constipation, n (%)175 (8.1)195 (6.1)0.006 b 0.168168 (7.8)160 (7.4)0.653 c 0.031SD, standard deviation; BMI, body mass index; BBPS, Boston Bowel Preparation Scale; PEG, polyethylene glycol; FIT, fecal immunochemical test; FOBT, fecal occult blood test; |SMD|, absolute standardized mean difference.aMann–Whitney U test.bChi-squared test.cMcNemar’s test.dWilcoxon signed-rank test.#Covariate balance was assessed using SMDs, calculated as Cohen’s d based on mean/standard deviations from raw data. For binary variables, standardized mean differences were calculated based on proportions (or percents) within each treatment group, using pooled standard deviations.@Sedation status was defined as propofol-based intravenous sedation versus no sedation.*Smoking and alcohol use were defined as ever use based on structured preprocedural assessment. Table 1.Baseline characteristics of the air insufflation and water exchange groups in the original and matched cohorts.Baseline characteristicsOriginal cohort (n=5,352)Matched cohort (n=4,322)Air insufflation (n =2,169)Water exchange (n =3,183)P Value|SMD| # Air insufflation (n =2,161)Water exchange (n =2,161)P Value|SMD| # Male, n (%)1,088 (50.2)1,620 (50.9)0.598 b 0.0151,085 (50.2)1,084 (50.2)1.000 c 0Age (years), mean (SD)59.0 (13.6)59.3 (12.3)0.910 a 0.01459.1 (13.6)59.3 (12.3)0.770 d 0.012BMI (kg/m2), mean (SD)25.3 (3.9)25.1 (3.7)0.240 a 0.03525.2 (3.9)25.2 (3.7)0.519 d 0.016Fully sedated colonoscopy, n (%) @ 1,379 (63.6)1,713 (53.8)<0.001 b 0.2241,371 (63.4)1,337 (61.9)0.107 c 0.035Bowel preparation with PEG, n (%)1,111 (51.2)1,796 (56.4)<0.001 b 0.1151,111 (51.4)1,110 (51.4)1.000 c 0Smoking, n (%)*218 (10.1)369 (11.6)0.076 b 0.086218 (10.1)230 (10.6)0.566 c 0.030Drinking, n (%)*202 (9.3)364 (11.4)0.013 b 0.125202 (9.3)210 (9.7)0.678 c 0.025FIT/FOBT positive, n (%)307 (14.2)510 (16.0)0.062 b 0.078307 (14.2)318 (14.7)0.658 c 0.022Previous abdominal surgery, n (%)330 (15.2)520 (16.3)0.270 b 0.046330 (15.3)342 (15.8)0.643 c 0.021Constipation, n (%)175 (8.1)195 (6.1)0.006 b 0.168168 (7.8)160 (7.4)0.653 c 0.031SD, standard deviation; BMI, body mass index; BBPS, Boston Bowel Preparation Scale; PEG, polyethylene glycol; FIT, fecal immunochemical test; FOBT, fecal occult blood test; |SMD|, absolute standardized mean difference.aMann–Whitney U test.bChi-squared test.cMcNemar’s test.dWilcoxon signed-rank test.#Covariate balance was assessed using SMDs, calculated as Cohen’s d based on mean/standard deviations from raw data. For binary variables, standardized mean differences were calculated based on proportions (or percents) within each treatment group, using pooled standard deviations.@Sedation status was defined as propofol-based intravenous sedation versus no sedation.*Smoking and alcohol use were defined as ever use based on structured preprocedural assessment. Baseline characteristics of the air insufflation and water exchange groups in the original and matched cohorts. SD, standard deviation; BMI, body mass index; BBPS, Boston Bowel Preparation Scale; PEG, polyethylene glycol; FIT, fecal immunochemical test; FOBT, fecal occult blood test; |SMD|, absolute standardized mean difference. aMann–Whitney U test. bChi-squared test. cMcNemar’s test. dWilcoxon signed-rank test. #Covariate balance was assessed using SMDs, calculated as Cohen’s d based on mean/standard deviations from raw data. For binary variables, standardized mean differences were calculated based on proportions (or percents) within each treatment group, using pooled standard deviations. @Sedation status was defined as propofol-based intravenous sedation versus no sedation. *Smoking and alcohol use were defined as ever use based on structured preprocedural assessment. Water exchange vs air insufflation: Procedure performance and bowel preparation In the matched cohort, cecal intubation was achieved in 100% of WE colonoscopies compared with 99.3% in the air-insufflation group. WE was associated with significantly longer procedure times across all measures. Compared with air insufflation, WE extended the mean cecal insertion time by 5.0 minutes, withdrawal time by 2.0 minutes, and total procedure time by 7.0 minutes (all P < 0.001; Table 2). Compared with air insufflation, WE significantly increased the mean total BBPS score by 1.1 points and improved the rate of adequate preparation by 9.0% (both P < 0.001; Table 2).Table 2.Procedure performance and detection outcomes by insertion method before and after propensity score matching.OutcomesOriginal cohort (n=5,352)Matched cohort (n=4,322)​Air insufflation (n =2,169)Water exchange (n =3,183)P valueAir insufflation (n =2,161)Water exchange (n =2,161)Effect Size @ (95% CI)NNT* (95% CI)P value Procedure performance Cecal intubation rate, % (n)99.3 (2,154)99.9 (3,181)<0.001 b 99.3 (2,146)100 (2,160)0.6 (0.3-1.0)167 (100-334)0.001 c Cecal insertion time (minutes), mean (SD)6.0 (4.5)11.0 (4.6)<0.001 a 6.0 (4.4)11.0 (4.6)5.0 (4.7-5.3)NA<0.001 d Withdrawal time (minutes), mean (SD)9.7 (6.0)11.8 (6.2)<0.001 a 9.6 (6.0)11.7 (6.1)2.0 (1.7-2.4)NA<0.001 d Total procedure time (minutes), mean (SD)15.7 (7.9)22.8 (8.2)<0.001 a 15.6 (7.9)22.6 (8.1)7.0 (6.5-7.5)NA<0.001 d Bowel preparation Total BBPS, mean (SD)6.5 (1.3)7.5 (1.0)<0.001 a 6.4 (1.3)7.5 (1.0)1.1 (1.0-1.2)NA<0.001 d Adequate bowel preparation rate, % (n)88.4 (1,918)97.4 (3,100)<0.001 b 88.4 (1,911)97.4 (2,105)9.0 (7.5-10.5)12 (10–14)<0.001 c Detection outcomes Adenoma detection rate, % (n)43.9 (953)58.3 (1,856)<0.001 b 44.0 (950)57.9 (1,252)14.0 (11.1-16.9)8 (6–10)<0.001 c Adenomas per colonoscopy, mean (SD)0.8 (1.2)1.1 (1.4)<0.001 a 0.8 (1.2)1.2 (1.4)0.4 (0.3-0.4)NA<0.001 d Adenomas per positive colonoscopy, mean (SD)1.2 (1.5)1.6 (1.6)0.001 a 1.2 (1.5)1.7 (1.6)0.4 (0.2-0.6)NA<0.001 d SSL detection rate, % (n)1.0 (21)1.9 (59)0.009 b 1.0 (21)1.9 (42)1.0 (0.3-1.7)100 (59–334)0.011 c Advanced ADR, % (n)8.3 (180)10.4 (332)0.009 b 8.3 (179)9.9 (214)1.6 (-0.1-3.3)NA0.072 c Polyp detection rate, % (n)50.6 (1,097)66.1 (2,103)<0.001 b 50.6 (1,093)65.4 (1,414)14.9 (12.0-17.7)7 (6–9)<0.001 c ADR was the prespecified primary outcome. Secondary, segmental, and component-specific outcomes were considered exploratory. No formal adjustment for multiple comparisons was performed; therefore, P values for these analyses should be interpreted as nominal.Abbreviations. ADR, adenoma detection rate; advanced ADR, advanced adenoma detection rate; BBPS, Boston Bowel Preparation Scale; NA, not applicable; NNT, number needed to treat; SD, standard deviation; SSL, sessile serrated lesion;aMann–Whitney U test.bChi-squared test.cMcNemar’s test.dWilcoxon signed-rank test.#Adequate bowel preparation: BBPS score ≥ 6 with each segment ≥ 2.@Effect sizes are presented as risk differences (RDs) in percentage points for binary outcomes and mean differences (MDs) for continuous outcomes, with corresponding 95% confidence intervals (CIs).*NNT was calculated as the reciprocal of the absolute risk difference for binary outcomes with statistically significant differences. NNT was not calculated for continuous outcomes or binary outcomes with confidence intervals crossing the null. Table 2.Procedure performance and detection outcomes by insertion method before and after propensity score matching.OutcomesOriginal cohort (n=5,352)Matched cohort (n=4,322)​Air insufflation (n =2,169)Water exchange (n =3,183)P valueAir insufflation (n =2,161)Water exchange (n =2,161)Effect Size @ (95% CI)NNT* (95% CI)P value Procedure performance Cecal intubation rate, % (n)99.3 (2,154)99.9 (3,181)<0.001 b 99.3 (2,146)100 (2,160)0.6 (0.3-1.0)167 (100-334)0.001 c Cecal insertion time (minutes), mean (SD)6.0 (4.5)11.0 (4.6)<0.001 a 6.0 (4.4)11.0 (4.6)5.0 (4.7-5.3)NA<0.001 d Withdrawal time (minutes), mean (SD)9.7 (6.0)11.8 (6.2)<0.001 a 9.6 (6.0)11.7 (6.1)2.0 (1.7-2.4)NA<0.001 d Total procedure time (minutes), mean (SD)15.7 (7.9)22.8 (8.2)<0.001 a 15.6 (7.9)22.6 (8.1)7.0 (6.5-7.5)NA<0.001 d Bowel preparation Total BBPS, mean (SD)6.5 (1.3)7.5 (1.0)<0.001 a 6.4 (1.3)7.5 (1.0)1.1 (1.0-1.2)NA<0.001 d Adequate bowel preparation rate, % (n)88.4 (1,918)97.4 (3,100)<0.001 b 88.4 (1,911)97.4 (2,105)9.0 (7.5-10.5)12 (10–14)<0.001 c Detection outcomes Adenoma detection rate, % (n)43.9 (953)58.3 (1,856)<0.001 b 44.0 (950)57.9 (1,252)14.0 (11.1-16.9)8 (6–10)<0.001 c Adenomas per colonoscopy, mean (SD)0.8 (1.2)1.1 (1.4)<0.001 a 0.8 (1.2)1.2 (1.4)0.4 (0.3-0.4)NA<0.001 d Adenomas per positive colonoscopy, mean (SD)1.2 (1.5)1.6 (1.6)0.001 a 1.2 (1.5)1.7 (1.6)0.4 (0.2-0.6)NA<0.001 d SSL detection rate, % (n)1.0 (21)1.9 (59)0.009 b 1.0 (21)1.9 (42)1.0 (0.3-1.7)100 (59–334)0.011 c Advanced ADR, % (n)8.3 (180)10.4 (332)0.009 b 8.3 (179)9.9 (214)1.6 (-0.1-3.3)NA0.072 c Polyp detection rate, % (n)50.6 (1,097)66.1 (2,103)<0.001 b 50.6 (1,093)65.4 (1,414)14.9 (12.0-17.7)7 (6–9)<0.001 c ADR was the prespecified primary outcome. Secondary, segmental, and component-specific outcomes were considered exploratory. No formal adjustment for multiple comparisons was performed; therefore, P values for these analyses should be interpreted as nominal.Abbreviations. ADR, adenoma detection rate; advanced ADR, advanced adenoma detection rate; BBPS, Boston Bowel Preparation Scale; NA, not applicable; NNT, number needed to treat; SD, standard deviation; SSL, sessile serrated lesion;aMann–Whitney U test.bChi-squared test.cMcNemar’s test.dWilcoxon signed-rank test.#Adequate bowel preparation: BBPS score ≥ 6 with each segment ≥ 2.@Effect sizes are presented as risk differences (RDs) in percentage points for binary outcomes and mean differences (MDs) for continuous outcomes, with corresponding 95% confidence intervals (CIs).*NNT was calculated as the reciprocal of the absolute risk difference for binary outcomes with statistically significant differences. NNT was not calculated for continuous outcomes or binary outcomes with confidence intervals crossing the null. Procedure performance and detection outcomes by insertion method before and after propensity score matching. ADR was the prespecified primary outcome. Secondary, segmental, and component-specific outcomes were considered exploratory. No formal adjustment for multiple comparisons was performed; therefore, P values for these analyses should be interpreted as nominal. Abbreviations. ADR, adenoma detection rate; advanced ADR, advanced adenoma detection rate; BBPS, Boston Bowel Preparation Scale; NA, not applicable; NNT, number needed to treat; SD, standard deviation; SSL, sessile serrated lesion; aMann–Whitney U test. bChi-squared test. cMcNemar’s test. dWilcoxon signed-rank test. #Adequate bowel preparation: BBPS score ≥ 6 with each segment ≥ 2. @Effect sizes are presented as risk differences (RDs) in percentage points for binary outcomes and mean differences (MDs) for continuous outcomes, with corresponding 95% confidence intervals (CIs). *NNT was calculated as the reciprocal of the absolute risk difference for binary outcomes with statistically significant differences. NNT was not calculated for continuous outcomes or binary outcomes with confidence intervals crossing the null. Detection performance In the matched cohort (Table 2), ADR was higher in the WE group than in the air-insufflation group (57.9% vs 44.0%; RD 14.0%, 95% CI 11.1–16.9; P<0.001), corresponding to an NNT of 8 (95% CI, 6–10). WE also significantly increased the PDR (RD 14.9%, 95% CI 12.0–17.6), APC (MD = 0.4; 95% CI 0.3–0.3), and APP (MD = 0.4; 95% CI 0.2–0.6). For SSL detection rate, the absolute increase was 1.0% (95% CI 0.3–1.7) but remained statistically significant (P=0.011). Conversely, advanced ADR was numerically higher with WE but did not achieve statistical significance (9.9% vs. 8.3%, P = 0.072). Regional impact on bowel preparation and detection rates Table 3 presents the regional outcomes for bowel preparation and lesion detection. WE consistently achieved significantly higher rates of adequate bowel preparation across all colonic segments compared with air insufflation (all P < 0.001, Table 3). Notably, for the right colon, the adequate preparation rate was 97.2% with WE versus 90.4% with air insufflation. For adenoma detection by segment, WE improved ADR across the colon (Table 3). WE significantly improved the right-sided ADR compared with air insufflation (31.8% vs. 18.0%, P < 0.001), corresponding to an NNT of 8 (95% CI 7 to 9). Higher ADRs were also consistently documented with WE in both the transverse and left colon (Table 3).Table 3.Segment-specific bowel preparation quality and detection outcomes.OutcomesOriginal cohort (n=5,352)Matched cohort (n=4,322)​Air insufflation (n =2,169)Water exchange (n =3,183)P valueAir insufflation (n =2,161)Water exchange (n =2,161)Effect Size @ (95% CI)NNT* (95% CI)P value Adequate bowel preparation rate, %(n) Right side colon BBPS ≥ 290.5 (1,946)97.2 (3,092)<0.001 b 90.4 (1,938)97.2 (2,100)6.7 (5.3-8.2)15 (13–19)<0.001 c Transverse colon BBPS ≥ 295.1 (2,052)99.4 (3,160)<0.001 b 95.1 (2,044)99.4 (2,146)4.2 (3.2-5.2)24 (20–32)<0.001 c Left side colon BBPS ≥ 294.8 (2,057)99.0 (3,151)<0.001 b 94.9 (2,050)98.9 (2,138)4.1 (3.0-5.1)25 (20–34)<0.001 c Adenoma detection rate, %(n) Right side colon ADR17.9 (389)31.9 (1,015)<0.001 b 18.0 (389)31.8 (687)13.8 (11.3-16.3)8 (7–9)<0.001 c Transverse colon ADR14.8 (320)19.4 (618)<0.001 b 14.7 (318)19.7 (426)5 (2.7-7.3)20 (14–38)<0.001 c Left side colon ADR29.5 (640)35.9 (1,144)<0.001 b 29.6 (639)35.6 (770)6.1 (3.3-8.8)17 (12–31)<0.001 c Advanced ADR, %(n) by segment Right side colon advanced ADR2.2 (47)3.5 (111)0.005 b 2.2 (47)3.3 (72)1.2 (0.2-2.1)84 (48–500)0.024 c Transverse colon advanced ADR0.8 (17)1.6 (52)0.007 b 0.8 (17)1.6 (34)0.8 (0.1-1.4)125 (72–1000)0.024 c Left side colon advanced ADR5.1 (110)6.1 (195)0.102 b 5 (108)5.7 (123)0.7 (-0.6-2.0)NA0.349 c Advanced ADR, %(n) by components Villous ADR %(n)4.1 (89)4.4 (139)0.639 b 4.1 (89)4.3 (94)0.2 (-1.0-1.4)NA0.761 c Size ≥ 10 mm ADR %(n)6.7 (145)9.1 (291)0.001 b 6.7 (144)8.7 (188)2.0 (0.5-3.6)50 (28–200)0.014 c High-grade dysplasia ADR %(n)1.1 (23)1.0 (32)0.845 b 1.1 (23)0.8 (17)-0.3 (-0.9-0.3)NA0.430 c SSL detection rate, %(n) Right side colon SSL detection rate0.4 (9)1.0 (32)0.015 b 0.4 (9)1.1 (24)0.7 (0.2-1.2)143 (84–500)0.014 c Transverse colon SSL detection rate0.1 (2)0.2 (7)0.363 b 0.1 (2)0.2 (4)0.1 (-0.1-0.3)NA0.687 c Left side colon SSL detection rate0.5 (11)0.8 (25)0.221 b 0.5 (11)0.8 (18)0.3 (-0.2-0.8)NA0.265 c ADR was the prespecified primary outcome. Secondary, segmental, and component-specific outcomes were considered exploratory. No formal adjustment for multiple comparisons was performed; therefore, P values for these analyses should be interpreted as nominal.Abbreviations. ADR, adenoma detection rate; advanced ADR, advanced adenoma detection rate; BBPS, Boston Bowel Preparation Scale; NA, not applicable; NNT, number needed to treat; SD, standard deviation; SSL, sessile serrated lesionAdvanced adenoma: Villous adenoma, ≥ 10 mm adenoma, High grade dysplasia.Adequate bowel preparation: BBPS score ≥ 6 with ≥ 2 in each segment.Right side colon: The right colon includes the cecum, the ascending colon and hepatic Flexure.Left side colon: The left colon includes splenic flexure, descending, sigmoid, and rectum.aMann–Whitney U test.bChi-squared test.cMcNemar’s test.dWilcoxon signed-rank test.@Effect sizes are presented as risk differences (RDs) in percentage points for binary outcomes and mean differences (MDs) for continuous outcomes, with corresponding 95% confidence intervals (CIs).*NNT was calculated as the reciprocal of the absolute risk difference for binary outcomes with statistically significant differences. NNT was not calculated for continuous outcomes or binary outcomes with confidence intervals crossing the null. Table 3.Segment-specific bowel preparation quality and detection outcomes.OutcomesOriginal cohort (n=5,352)Matched cohort (n=4,322)​Air insufflation (n =2,169)Water exchange (n =3,183)P valueAir insufflation (n =2,161)Water exchange (n =2,161)Effect Size @ (95% CI)NNT* (95% CI)P value Adequate bowel preparation rate, %(n) Right side colon BBPS ≥ 290.5 (1,946)97.2 (3,092)<0.001 b 90.4 (1,938)97.2 (2,100)6.7 (5.3-8.2)15 (13–19)<0.001 c Transverse colon BBPS ≥ 295.1 (2,052)99.4 (3,160)<0.001 b 95.1 (2,044)99.4 (2,146)4.2 (3.2-5.2)24 (20–32)<0.001 c Left side colon BBPS ≥ 294.8 (2,057)99.0 (3,151)<0.001 b 94.9 (2,050)98.9 (2,138)4.1 (3.0-5.1)25 (20–34)<0.001 c Adenoma detection rate, %(n) Right side colon ADR17.9 (389)31.9 (1,015)<0.001 b 18.0 (389)31.8 (687)13.8 (11.3-16.3)8 (7–9)<0.001 c Transverse colon ADR14.8 (320)19.4 (618)<0.001 b 14.7 (318)19.7 (426)5 (2.7-7.3)20 (14–38)<0.001 c Left side colon ADR29.5 (640)35.9 (1,144)<0.001 b 29.6 (639)35.6 (770)6.1 (3.3-8.8)17 (12–31)<0.001 c Advanced ADR, %(n) by segment Right side colon advanced ADR2.2 (47)3.5 (111)0.005 b 2.2 (47)3.3 (72)1.2 (0.2-2.1)84 (48–500)0.024 c Transverse colon advanced ADR0.8 (17)1.6 (52)0.007 b 0.8 (17)1.6 (34)0.8 (0.1-1.4)125 (72–1000)0.024 c Left side colon advanced ADR5.1 (110)6.1 (195)0.102 b 5 (108)5.7 (123)0.7 (-0.6-2.0)NA0.349 c Advanced ADR, %(n) by components Villous ADR %(n)4.1 (89)4.4 (139)0.639 b 4.1 (89)4.3 (94)0.2 (-1.0-1.4)NA0.761 c Size ≥ 10 mm ADR %(n)6.7 (145)9.1 (291)0.001 b 6.7 (144)8.7 (188)2.0 (0.5-3.6)50 (28–200)0.014 c High-grade dysplasia ADR %(n)1.1 (23)1.0 (32)0.845 b 1.1 (23)0.8 (17)-0.3 (-0.9-0.3)NA0.430 c SSL detection rate, %(n) Right side colon SSL detection rate0.4 (9)1.0 (32)0.015 b 0.4 (9)1.1 (24)0.7 (0.2-1.2)143 (84–500)0.014 c Transverse colon SSL detection rate0.1 (2)0.2 (7)0.363 b 0.1 (2)0.2 (4)0.1 (-0.1-0.3)NA0.687 c Left side colon SSL detection rate0.5 (11)0.8 (25)0.221 b 0.5 (11)0.8 (18)0.3 (-0.2-0.8)NA0.265 c ADR was the prespecified primary outcome. Secondary, segmental, and component-specific outcomes were considered exploratory. No formal adjustment for multiple comparisons was performed; therefore, P values for these analyses should be interpreted as nominal.Abbreviations. ADR, adenoma detection rate; advanced ADR, advanced adenoma detection rate; BBPS, Boston Bowel Preparation Scale; NA, not applicable; NNT, number needed to treat; SD, standard deviation; SSL, sessile serrated lesionAdvanced adenoma: Villous adenoma, ≥ 10 mm adenoma, High grade dysplasia.Adequate bowel preparation: BBPS score ≥ 6 with ≥ 2 in each segment.Right side colon: The right colon includes the cecum, the ascending colon and hepatic Flexure.Left side colon: The left colon includes splenic flexure, descending, sigmoid, and rectum.aMann–Whitney U test.bChi-squared test.cMcNemar’s test.dWilcoxon signed-rank test.@Effect sizes are presented as risk differences (RDs) in percentage points for binary outcomes and mean differences (MDs) for continuous outcomes, with corresponding 95% confidence intervals (CIs).*NNT was calculated as the reciprocal of the absolute risk difference for binary outcomes with statistically significant differences. NNT was not calculated for continuous outcomes or binary outcomes with confidence intervals crossing the null. Segment-specific bowel preparation quality and detection outcomes. ADR was the prespecified primary outcome. Secondary, segmental, and component-specific outcomes were considered exploratory. No formal adjustment for multiple comparisons was performed; therefore, P values for these analyses should be interpreted as nominal. Abbreviations. ADR, adenoma detection rate; advanced ADR, advanced adenoma detection rate; BBPS, Boston Bowel Preparation Scale; NA, not applicable; NNT, number needed to treat; SD, standard deviation; SSL, sessile serrated lesion Advanced adenoma: Villous adenoma, ≥ 10 mm adenoma, High grade dysplasia. Adequate bowel preparation: BBPS score ≥ 6 with ≥ 2 in each segment. Right side colon: The right colon includes the cecum, the ascending colon and hepatic Flexure. Left side colon: The left colon includes splenic flexure, descending, sigmoid, and rectum. aMann–Whitney U test. bChi-squared test. cMcNemar’s test. dWilcoxon signed-rank test. @Effect sizes are presented as risk differences (RDs) in percentage points for binary outcomes and mean differences (MDs) for continuous outcomes, with corresponding 95% confidence intervals (CIs). *NNT was calculated as the reciprocal of the absolute risk difference for binary outcomes with statistically significant differences. NNT was not calculated for continuous outcomes or binary outcomes with confidence intervals crossing the null. The subgroup analysis regarding the impact on advanced ADR and SSL detection is shown in Table 3. WE significantly increased advanced ADRs in the right (+1.2%, P = 0.024) and transverse colon (+0.8%, P = 0.024), but not in the left colon (P = 0.349; Table 3). Analyzing advanced components revealed that WE enhanced the detection of large adenomas (≥1.0 cm) by 2.0% (P = 0.014), while villous adenoma and high-grade dysplasia detection rates were statistically similar between the two groups. For SSLs, WE significantly increased the right-sided detection rate (+0.7%, P = 0.014) compared with air insufflation. Conversely, SSL detection rates in the transverse and left colon showed no statistically significant differences (both P > 0.05). Factors associated with ADR To identify the independent predictors of ADR, a multivariable logistic regression analysis with forward stepwise selection was performed (Table 4). After adjusting for potential confounding factors, the water exchange method remained a significant independent predictor of a higher ADR compared with air insufflation (adjusted odds ratio = 1.551; 95% CI: 1.279–1.882; P < 0.001). Furthermore, male gender, older age (≥65 years), smoking, positive FIT/FOBT status, and a prolonged withdrawal time (≥9 minutes) were identified as significant independent predictors of an increased ADR. Conversely, endoscopist YHH was independently associated with a lower ADR compared with endoscopist CWT.Table 4.Univariate and multivariate logistic regression analyses of factors associated with adenoma detection rate (ADR) (Matched cohort; n=4,322).VariableADR (%)Univariate logistic regressionMultivariate logistic regression#OR (95% CI)P valueaOR (95% CI)P value Insertion method Air44.0Ref—Ref— Water57.91.756 (1.556, 1.981)<0.0011.551 (1.279-1.882)<0.001 Gender Female45.8Ref—Ref— Male56.11.516 (1.345, 1.709)<0.0011.285 (1.117-1.479)<0.001 Age (years old) < 65 years old44.2Ref—Ref— ≥ 65 years old61.42.004 (1.769, 2.270)<0.0011.951 (1.697-2.242)<0.001 BMI (kg/m 2 ) <2448.1Ref——— ≥2452.81.203 (1.065, 1.360)0.003—— Fully sedated colonoscopy @ Non- sedated49.8Ref——— Fully sedated51.71.079 (0.954,1.221)0.225—— Preparation drug Non-PEG49.1Ref——— PEG52.71.157 (1.027, 1.304)0.016—— Smoking * No49.6Ref—Ref​ Yes62.31.675 (1.370, 2.048)<0.0011.274 (1.009-1.608)0.042 Drinking * No50.2Ref——— Yes57.81.355 (1.104, 1.664)0.004—— FIT/FOBT positive No50.7Ref—Ref​ Yes64.11.796 (1.508, 2.140)<0.0011.568 (1.290-1.907)<0.001 Previous abdominal surgery No56.6Ref——— Yes57.91.117 (0.947, 1.317)0.190—— Constipation No53.4Ref——— Yes52.90.906 (0.757, 1.086)0.285—— Adequate bowel preparation & No46.1Ref——— Yes51.31.234 (0.977, 1.558)0.078—— Withdrawal time (min) <926.0Ref—Ref— ≥969.56.482 (5.666, 7.416)<0.0016.161 (5.311- 7.146)<0.001 Endoscopist CWT43.8Ref—Ref— YHH54.72.235 (1.876, 2.663)<0.0010.814 (0.664, 0.997)0.046 CPT63.51.548 (1.355, 1.767)<0.0011.233 (0.979, 1.553)0.076BBPS, Boston Bowel Preparation Scale; BMI, body mass index; CI, confidence interval; FIT, fecal immunochemical test; FOBT, fecal occult blood test; OR, odds ratio; aOR, adjusted odds ratio; PEG, polyethylene glycol.&Adequate bowel preparation was defined as a total BBPS score ≥6 with each segment score ≥2.*Smoking and alcohol use were defined as ever use based on structured preprocedural assessment.@Sedation status was defined as propofol-based intravenous sedation versus no sedation.#Multivariable model adjusted for insertion method, gender, and age (forced entry), and variables with univariable p < 0.05 (BMI, preparation drug, smoking, drinking, FIT/FOBT positive, withdrawal time, and endoscopist) via forward stepwise selection (entry: P < 0.05, removal: P > 0.10). Table 4.Univariate and multivariate logistic regression analyses of factors associated with adenoma detection rate (ADR) (Matched cohort; n=4,322).VariableADR (%)Univariate logistic regressionMultivariate logistic regression#OR (95% CI)P valueaOR (95% CI)P value Insertion method Air44.0Ref—Ref— Water57.91.756 (1.556, 1.981)<0.0011.551 (1.279-1.882)<0.001 Gender Female45.8Ref—Ref— Male56.11.516 (1.345, 1.709)<0.0011.285 (1.117-1.479)<0.001 Age (years old) < 65 years old44.2Ref—Ref— ≥ 65 years old61.42.004 (1.769, 2.270)<0.0011.951 (1.697-2.242)<0.001 BMI (kg/m 2 ) <2448.1Ref——— ≥2452.81.203 (1.065, 1.360)0.003—— Fully sedated colonoscopy @ Non- sedated49.8Ref——— Fully sedated51.71.079 (0.954,1.221)0.225—— Preparation drug Non-PEG49.1Ref——— PEG52.71.157 (1.027, 1.304)0.016—— Smoking * No49.6Ref—Ref​ Yes62.31.675 (1.370, 2.048)<0.0011.274 (1.009-1.608)0.042 Drinking * No50.2Ref——— Yes57.81.355 (1.104, 1.664)0.004—— FIT/FOBT positive No50.7Ref—Ref​ Yes64.11.796 (1.508, 2.140)<0.0011.568 (1.290-1.907)<0.001 Previous abdominal surgery No56.6Ref——— Yes57.91.117 (0.947, 1.317)0.190—— Constipation No53.4Ref——— Yes52.90.906 (0.757, 1.086)0.285—— Adequate bowel preparation & No46.1Ref——— Yes51.31.234 (0.977, 1.558)0.078—— Withdrawal time (min) <926.0Ref—Ref— ≥969.56.482 (5.666, 7.416)<0.0016.161 (5.311- 7.146)<0.001 Endoscopist CWT43.8Ref—Ref— YHH54.72.235 (1.876, 2.663)<0.0010.814 (0.664, 0.997)0.046 CPT63.51.548 (1.355, 1.767)<0.0011.233 (0.979, 1.553)0.076BBPS, Boston Bowel Preparation Scale; BMI, body mass index; CI, confidence interval; FIT, fecal immunochemical test; FOBT, fecal occult blood test; OR, odds ratio; aOR, adjusted odds ratio; PEG, polyethylene glycol.&Adequate bowel preparation was defined as a total BBPS score ≥6 with each segment score ≥2.*Smoking and alcohol use were defined as ever use based on structured preprocedural assessment.@Sedation status was defined as propofol-based intravenous sedation versus no sedation.#Multivariable model adjusted for insertion method, gender, and age (forced entry), and variables with univariable p < 0.05 (BMI, preparation drug, smoking, drinking, FIT/FOBT positive, withdrawal time, and endoscopist) via forward stepwise selection (entry: P < 0.05, removal: P > 0.10). Univariate and multivariate logistic regression analyses of factors associated with adenoma detection rate (ADR) (Matched cohort; n=4,322). BBPS, Boston Bowel Preparation Scale; BMI, body mass index; CI, confidence interval; FIT, fecal immunochemical test; FOBT, fecal occult blood test; OR, odds ratio; aOR, adjusted odds ratio; PEG, polyethylene glycol. &Adequate bowel preparation was defined as a total BBPS score ≥6 with each segment score ≥2. *Smoking and alcohol use were defined as ever use based on structured preprocedural assessment. @Sedation status was defined as propofol-based intravenous sedation versus no sedation. #Multivariable model adjusted for insertion method, gender, and age (forced entry), and variables with univariable p < 0.05 (BMI, preparation drug, smoking, drinking, FIT/FOBT positive, withdrawal time, and endoscopist) via forward stepwise selection (entry: P < 0.05, removal: P > 0.10). Discussion In this real-world, propensity score-matched cohort, WE colonoscopy was associated with higher neoplasia detection compared with air insufflation. A clinically relevant increase in ADR was observed, accompanied by higher adenoma per colonoscopy and greater right-sided detection. This right-sided association is clinically relevant because proximal lesions, particularly sessile serrated lesions, are more easily missed. However, the higher detection observed with WE was accompanied by longer procedure time, highlighting an important workflow trade-off when considering wider implementation of WE. Randomized trials and meta-analyses have demonstrated that WE improves overall ADR and may enhance the detection of advanced and serrated lesions compared with standard gas insufflation techniques.13,16–18,31 A network meta-analysis by Fuccio et al. reported an approximately 11% absolute increase in ADR with WE compared with air insufflation (41.7% vs. 30.2%). 17 It is noteworthy that the baseline detection performance in our study was high. Even the air insufflation group achieved an ADR of 44.0%, which exceeds the pooled ADR of 26.5% reported in large real-world population studies. 21 Against this high background, WE achieved a substantial absolute ADR increase of 14.0 percentage points (57.9% vs. 44.0%; P < 0.001), with a corresponding NNT of 8. Our real-world findings extend this evidence by demonstrating comparable benefits under routine clinical conditions within a broad practice population. Additionally, the mean APC significantly increased in the WE group (1.2 vs. 0.8; P < 0.001), a figure that exceeds the current minimum guideline of 0.6. 25 Since APC is often considered a more linear reflection of inspection quality than the binary ADR, 25 this finding suggests that WE effectively exposes multiple lesions that might otherwise be obscured. The right colon remains an area historically prone to missed lesions, with pooled adenoma miss rates reaching approximately 26% largely due to small or flat lesions. 4 A potential advantage of WE is its ability to improve visualization in this challenging segment. Segmental analysis showed that the absolute increase in right-sided ADR (+13.8%) substantially exceeded gains in the transverse (+5.0%) and left colon (+6.1%). This suggests that WE may improve visualization in the right colon, where subtle lesions are frequent and residual fluid may obscure the view. This benefit aligns with a systematic review reporting a significant risk ratio of 1.43 favoring WE for adenoma detection in the right colon. 22 Together, these findings are consistent with the clinical relevance of WE for proximal colon inspection in routine practice. Regarding SSLs, we observed a modest but statistically significant increase in overall SSL detection (1.9% vs. 1.0%; RD 1.0%, 95% CI 0.3–1.7; P = 0.011). A similar increase was observed in the right colon (1.1% vs. 0.4%; P = 0.014). Although the low event rates warrant cautious interpretation, these findings are consistent with a recent network meta-analysis of 11 randomized controlled trials, which found that WE improved serrated lesion detection compared with air insufflation and auxiliary devices. 32 Similarly, Cheng et al. reported that WE significantly reduced the miss rate for right-sided sessile polyps compared with CO2 insufflation. 14 Collectively, these data support the hypothesis that the water exchange mechanism effectively washes away the mucus caps that typically camouflage proximal serrated lesions. 33 Given that every 1% increase in proximal serrated polyp detection correlates with a 7% reduction in PCCRC risk, 5 this improved detection holds significant clinical value. Previous pooled analyses and meta-analyses have established that WE significantly increases advanced ADR, with risk ratios around 1.38, outperforming auxiliary devices like Endocuff.19,34 Although the overall advanced ADR in our study was numerically higher in the WE group (9.9% vs. 8.3%), this difference did not reach statistical significance (P = 0.072). This study was powered for the primary endpoint of overall ADR, but not specifically for secondary endpoints with lower event rates, such as advanced ADR. The modest absolute difference in advanced ADR and the CI crossing the null suggest that this endpoint may have been underpowered. However, the study showed a significant association between WE and size-based advanced lesion detection. WE significantly increased the detection of adenomas ≥ 10 mm (8.7% vs. 6.7%; P = 0.014). Although the overall advanced ADR did not reach statistical significance, segment-level analyses showed statistically significant increases in advanced ADR in the right and transverse colon. These findings suggest that WE may not merely increase the detection of diminutive polyps. Rather, the significant increase in adenomas ≥10 mm supports a potential benefit for clinically relevant size-based outcomes, although this finding should be interpreted cautiously as a secondary outcome. The improved detection rates observed in this study are mechanistically supported by superior bowel preparation. By design, WE achieves luminal content exchange during insertion, resulting in a cleaner mucosal surface for withdrawal inspection. This “salvage cleaning” effect was evident in our data. The WE group demonstrated significantly higher median Total BBPS scores (7.5 vs. 6.4; P < 0.001) and higher rates of adequate preparation (97.4% vs. 88.4%; P < 0.001) compared to the air insufflation group. Crucially, this benefit was most pronounced in the right colon (97.2% vs. 90.4%; P < 0.001), where WE effectively converts adherent debris into a suctionable suspension. This improvement in right-colon adequacy likely drives the observed gains in right-sided ADR. Our findings are consistent with Fuccio et al., who reported superior BBPS scores with WE in a large meta-analysis, 17 and other randomized controlled trials showing higher rates of excellent cleanliness with water exchange.15,16,31 By salvaging suboptimal preparations intra-procedurally, WE may reduce the need for early repeat examinations and lower the associated patient and healthcare burden. While WE was associated with a longer median cecal insertion time compared to air insufflation (11 min vs. 6 min; P < 0.001), this increase was associated with a 100% cecal intubation rate and significantly higher diagnostic yields for proximal and advanced lesions. Our real-world data showed a total procedure time increase of 7.0 minutes (22.6 min vs. 15.6 min). Although the increase in total procedure time exceeds the 2–4-minute difference reported in previous meta-analyses. 20 the 7-minute investment may reduce missed lesions and the need for earlier surveillance. Consequently, the technique is feasible for routine practice, especially when balanced against the substantial gains in detection. Efficiency may be further optimized through strategies such as water volume discipline, scope shortening maneuvers, and adjuncts like cap assistance.28,35 Additionally, patient selection represents a practical lever to maintain procedural efficiency. Prior studies suggest that female sex, lower BMI, constipation, and inadequate bowel preparation are linked to prolonged insertion time with the water-exchange technique. 28 Endoscopists in the training phase may utilize these profiles to select appropriate candidates, ensuring that the clinical benefits of WE are achieved without compromising operational efficiency. Some limitations of this study should be mentioned. First, in this retrospective single-center analysis, group allocation depended on patient preference and self-payment. Although we employed PSM to balance key baseline characteristics, this method only addresses measured covariates within the propensity model. Therefore, residual confounding from unmeasured factors, such as socioeconomic status, health awareness, or preventive care motivation, cannot be entirely excluded, and our findings demonstrate associations rather than direct causality. Additionally, while the superior bowel preparation quality in the WE group reflects an intra-procedural salvage cleaning effect rather than a baseline imbalance, it remains a potential confounder. Although baseline regimens were balanced via PSM and a post hoc analysis showed no significant association between adequate preparation and ADR in our matched cohort, residual confounding from unmeasured factors cannot be ruled out. Second, the inclusion of only three endoscopists limits the generalizability of our findings to real-world settings with varying levels of operator expertise. Because all procedures were performed by highly experienced WE endoscopists in a high-volume center, the observed benefits and time trade-offs may not be directly generalizable to lower-volume centers or less experienced operators. Consequently, operational outcomes may vary depending on institutional workflow constraints and external validation through multicenter studies is required. Third, the prolonged withdrawal time in the WE group may have contributed to improved detection. However, the technique remained an independent predictor of a higher ADR after adjusting for withdrawal time in a multivariable analysis, indicating that the mucosal cleansing benefits of WE independently drive the increased diagnostic yield. Fourth, endoscopists were not blinded to the insertion method because of the retrospective real-world study design, which may have introduced detection or performance bias. Fifth, although sedation status was balanced using PSM and adjusted for in the regression models, individual sedation depth and total propofol doses were not analyzed, meaning residual confounding related to sedation intensity cannot be entirely excluded. Sixth, multiple secondary and segmental comparisons were exploratory and lacked formal multiplicity adjustments. These findings should be interpreted cautiously as supportive rather than confirmatory. Furthermore, while the large real-world sample size yielded statistical significance for certain secondary outcomes, their small absolute risk differences warrant caution, as statistical significance does not inherently imply clinical relevance. Finally, we did not evaluate long-term outcomes, such as interval cancer rates or cost-effectiveness, which are essential to quantify the full clinical and economic impact of adopting WE at scale. Conclusions In this large-scale, propensity score-matched analysis, WE was associated with higher adenoma detection than air insufflation in a real-world setting. An important finding of this study is that the magnitude of benefit (ADR +14.0%) persisted even against a high-performance baseline, where the air insufflation group’s ADR (44.0%) already substantially exceeded international benchmarks. Furthermore, our data suggest that the association between WE and improved detection was not limited to diminutive polyps. We observed significant increases in the detection of adenomas ≥ 10 mm and proximal SSLs. These gains may be partly related to superior bowel cleansing, particularly in the right colon. Although WE required a median investment of 7 additional minutes, it was associated with a 100% cecal intubation rate and higher diagnostic yield, validating its feasibility and value as a standard insertion technique in routine clinical practice. These findings should be interpreted considering the retrospective, single-center design and require confirmation in prospective multicenter studies. Supplemental material Supplemental material - Real-world evidence of water exchange colonoscopy improving adenoma detection rates: A propensity score-matched analysis Supplemental material for Real-world evidence of water exchange colonoscopy improving adenoma detection rates: A propensity score-matched analysis by Ying-Ju Chao, Yu-Hsi Hsieh, Chia-Pei Tang, Ping-Hung Ko, Yen-Chun Chen, Chia-Hsin Cheng, Chih-Wei Tseng in Therapeutic Advances in Gastroenterology.