Colonoscopy Quality and Strategies for Improvement.

INTRODUCTION
Colorectal cancer (CRC) is one of the most prevalent cancers worldwide, with nearly 2 million new cases and 1 million deaths reported in 2020.1 The incidence of CRC has been steadily increasing in Korea, where it ranks as the second most common cancer diagnosed and the third leading cause of cancer deaths,2 thus emerging as a significant public health concern. Colonoscopy serves as a cornerstone screening method for CRC. As most CRC originates from adenomas, and it typically takes many years for adenomas to progress to cancer, detection and removal of adenomas through colonoscopy can reduce the incidence of CRC by up to 76%–90% and prevent CRC mortality by 53%.3,4 The NordICC study, the first randomized controlled trial (RCT) of invitation to colonoscopy versus usual care, demonstrated the effectiveness of screening colonoscopy, reporting a lower 10-year CRC risk among participants invited to screening, with the prespecified ascertainment of CRC mortality at 15 years still pending.5 Decision and cost-effectiveness analyses also predict colonoscopy to be the most effective method for prevention of CRC incidence,6 assuming uptake by screen-eligible persons, further supporting its role in CRC screening.
However, colonoscopy is imperfect and highly dependent on colonoscopist performance. Colonoscopy misses approximately 26% of adenomas, 9% of advanced adenomas, and 27% of serrated polyps,7 with detection rates varying among colonoscopists.8,9 Furthermore, lesions missed during colonoscopy are believed to be responsible for 50% to 60% of post-colonoscopy CRC (PCCRC), which occurs in the interval between scheduled colonoscopies10 and contributes to approximately 1.8% to 9% of all CRC cases.11 As Sidney Winawer has written, CRC screening is evidence-based but resource-driven, and the best screening modality is not only the one that gets done, but also done well.12
Colonoscopy can serve as a primary screening method or a diagnostic tool after other positive screening tests. In Korea, the National Cancer Screening Program has offered annual fecal immunochemical test (FIT) as the primary CRC screening tool, followed by confirmative colonoscopy for individuals with positive FIT results since 2004.13 Therefore, colonoscopy performed with high quality is essential to achieve the anticipated screening benefits, regardless of the screening strategy a country adopts.14 In response to the need for high-quality colonoscopy, colonoscopy quality indicators are introduced and updated by organizations in various countries.15-22 This review highlights the recent updates of colonoscopy quality indicators and presents strategies for improving colonoscopy quality.

PROCESS INDICATORS
Bowel preparation is a critical component of colonoscopy quality and is closely linked to other quality measures. Two key indicators–cecal intubation rate (CIR) and adenoma detection rate (ADR)–are associated with the bowel preparation adequacy.23,24 Inadequate bowel preparation increases costs, adverse events, and the need for repeat procedures at accelerated intervals, and more importantly, increases the risk of PCCRC.25,26 A landmark study on the impact of suboptimal bowel preparation revealed that the overall adenoma miss rate was as high as 42%, with the miss rate for advanced adenoma at 27%.27 A meta-analysis reported that ADR and advanced-ADR were significantly higher with adequate preparation compared to inadequate preparation (odds ratio [OR], 1.30; 95% confidence interval [CI], 1.19 to 1.42 and OR, 1.30; 95% CI, 1.02 to 1.67, respectively).24 Additionally, a Spanish observational study on PCCRC characteristics found that among 1,997 PCCRC cases (10-year rate of 6.7%), 36.3% had inadequate bowel preparation.26
Current guidelines emphasize documenting bowel preparation quality and recommend an early repeat colonoscopy within 1 year when bowel preparation for a screening or surveillance colonoscopy is deemed inadequate, defined as the inability to identify lesions >5 mm.28 The updated American College of Gastroenterology (ACG)/American Society for Gastrointestinal Endoscopy (ASGE) guidelines set a 90% threshold for bowel preparation adequacy.22 Similarly, the American Gastroenterological Association (AGA) clinical practice update recommends a minimum target of 90%, with an aspirational target of 95%,21 and the European Society of Gastrointestinal Endoscopy (ESGE) guidelines suggest a minimum standard of 90%, with a target standard of 95%.19 Various scoring systems are used to assess the bowel preparation adequacy: the Aronchick Scale of excellent, good, and fair; the Ottawa Scale ≤7; and the Boston Bowel Preparation Scale ≥2 out of 3 in all three colon segments.29-31 In clinical practice, bowel preparation quality exists on a spectrum, and uncertainty persists regarding management for patients who fall in the intermediate range. Patients with “fair” preparation are often recommended a shortened surveillance interval, with increased variability in recommendations among endoscopists.32 It is noteworthy that a meta-analysis found no significant difference in ADR between indeterminate (fair) and high-quality (excellent or good) cleansing.24 Whatever scoring system is used, the primary focus for endoscopists should be to assess the quality of bowel preparation based on their ability to identify polyps after clearing retained fluid or stool and to recommend the timing of the next colonoscopy accordingly.
Despite the importance of bowel cleanliness, suboptimal bowel preparation occurs in up to 20%–30% of colonoscopies, imposing a considerable systemic burden.33 Because optimal performance depends largely on patient adherence to the preparation process, recent studies have focused on strategies to improve compliance. Traditionally, patients were restricted to clear liquids the day before colonoscopy and prohibited from certain foods for several days. However, recent evidence supports a simpler, patient-centered approach using low-residue and low-fiber diets with only a 1-day restriction.34,35 Moreover, the availability of low-volume bowel preparation regimens has proven essential to improve bowel preparation quality. A meta-analysis indicates that low- and high-volume regimens achieve comparable cleansing efficacy when administered in split doses, but tolerability is superior with low-volume regimens.36
Cecal intubation is indispensable for a complete colonoscopy, and it therefore serves as a foundation of high-quality colonoscopy. Cecal intubation is defined as passage of the colonoscope tip proximal to the ileocecal valve and fully into the cecal caput, allowing identification of the appendiceal orifice and medial wall of the cecum on the side of the ileocecal valve.17 Beyond its role as a marker of competence during colonoscopy training, cecal intubation is also a key indicator of colonoscopy quality: the CIR is inversely associated with both ADR and PCCRC. In the National Health Service Bowel Cancer Screening Programme, despite the mean CIR of 95.2%, the CIR varied from 76.2% to 100%, and correlated positively with ADR (Spearman rank correlation coefficient, 0.023).37 An Ontario Cancer Registry study reported that patients undergoing colonoscopy performed by an endoscopist with a high CIR (≥95%) were less likely to develop PCCRC than those with a low CIR (<80%) (OR, 0.72; 95% CI, 0.53 to 0.97 for proximal cancers and OR, 0.73; 95% CI, 0.54 to 0.97 for distal cancers, respectively).38 Accordingly, the recent ACG/ASGE guidelines recommend a performance target of ≥95%.22 Likewise, the AGA clinical practice update and ESGE guidelines suggest a minimum target of 90%, with an aspirational target of 95%, applicable to all screening, surveillance, or diagnostic colonoscopies.19,21 To ensure high-quality colonoscopy, cecal landmarks, including appendiceal orifice and ileocecal valve, should be documented in written reports and with photodocumentation. Meticulous cecal image documentation is associated with a higher polyp detection rate (PDR) (OR, 2.53; 95% CI, 1.45 to 3.59), further underscoring its association with colonoscopy quality.39 In clinical practice, adequate bowel preparation–a prerequisite for achieving cecal intubation–is essential. Understanding factors contributing to failure to reach the cecum, such as angulation, redundant colon, loop formation, pain and discomfort, and underlying pathologic processes, can be beneficial. Intubation tools, including variable stiffness insertion tubes and pediatric colonoscopies, as well as techniques like water-aided colonoscopy, can help low performers improve their CIR.40,41 Once consistent high-level performance has been demonstrated, the CIR can be measured intermittently or omitted, as noted in the latest ACG/ASGE guidelines.22
As mucosal inspection occurs primarily during withdrawal, an appropriate withdrawal time (WT) is another essential component of high-quality colonoscopy. WT is defined as the time from reaching the appendiceal orifice to the completion of retroflexion in the rectum during normal colonoscopies without biopsy or other intervention.22 Dedicating adequate time to the withdrawal phase in screening colonoscopy is strongly associated with improved neoplasia detection. A landmark study demonstrated that colonoscopists with a mean WT ≥6 minutes had significantly higher ADR compared to those with WT <6 minutes (28.3% vs 11.8%).42 Additionally, data from the Minnesota Cancer Surveillance System indicated an inverse relationship between a physician’s mean annual WT and the risk of PCCRC.43 Emerging evidence suggests that both ADR and the detection of serrated lesions improve with each additional minute of WT beyond 6 minutes, with a plateau in benefit at approximately 9 minutes.44 A multicenter RCT further supported this, reporting a significant improvement in ADR when WT was prolonged from 6 to 9 minutes.45 The AGA clinical practice update and ESGE guidelines recommended a minimum WT target of 6 minutes, with an aspirational target of 9 minutes (AGA) and a target standard of 10 minutes (ESGE).19,21 Incorporating the latest data, the ACG/ASGE guidelines recently set an 8-minute average performance target for WT.22 However, it is important to note that simply increasing WT is not inherently meaningful; it contributes to high-quality colonoscopy only when monitored in conjunction with ADR. The most critical factor remains conducting a careful and detailed inspection, involving withdrawal techniques such as adequate insufflation, thorough examination of flexures and proximal sides of haustral folds, and suctioning of residual liquid, all of which require sufficient time during instrument withdrawal.46,47 Notably, if colonoscopists with low ADR also have shorter WT, refining their inspection techniques could significantly enhance overall colonoscopy quality as WT increases in order to allow adequate inspection.

DETECTION INDICATORS

1. Adenoma detection rate
Given the fundamental role of detecting and treating precancerous lesions through colonoscopy in preventing CRC, the ADR is widely regarded as the core quality indicator in colonoscopy. ADR is defined as the percentage of patients undergoing colonoscopy in whom one or more adenomas are resected and verified by pathology, with the original definition restricted to screening colonoscopies.17 It is recognized as the most reliable predictor of PCCRC risk. A Polish study involving 45,260 screening colonoscopies demonstrated a significant association between an endoscopist’s ADR and the risk of PCCRC, with hazard ratios (HRs) more than 10 times higher when colonoscopies were performed by endoscopists with ADRs <20% compared to those of ADRs ≥20% (HRs, 10.94, 10.75, 12.50 for ADRs <11.0%, 11.0%–14.9%, 15.0%–19.9%, respectively).48 Additionally, a large community-based United States population study involving 314,872 colonoscopies reinforced these findings, showing a strong inverse relationship between ADR and PCCRC risk.49 Patients of physicians in the highest ADR quintile (33.5% to 52.5%) had a significantly reduced PCCRC risk (adjusted HR, 0.52; 95% CI, 0.39 to 0.69) compared to those in the lowest ADR quintile (7.4% to 19.0%). Importantly, the association was approximately linear across ADR ranges, with each 1.0% increase in ADR associated with a 3.0% reduction in PCCRC risk. ADR holds equal importance in FIT-based CRC screening programs. An Italian FIT-based CRC screening program reported a significant inverse relationship between ADR and PCCRC risk, with a 2.35-fold higher risk observed in the lowest ADR group (20% to 39.9%) compared to the highest (55% to 70%).50 Most recently, an association has been reported between ADR and prevalent CRC detection, suggesting that endoscopists with lower ADRs may be missing prevalent CRCs more often than endoscopists with higher ADR.51 Consequently, it is imperative to measure the ADRs of all colonoscopists and to support those with low ADRs in achieving acceptable performance.
The previous ACG/ASGE guidelines recommend a minimum ADR target of 25% for patients aged >50 years undergoing screening colonoscopy (30% in men and 20% in women).17 Similarly, the ESGE guidelines recommend a target of ADR ≥25%.19 The GI Quality Improvement Consortium (GIQuIC) registry showed a significant increase in ADR over time, reaching 38.1% in 2018.52 A recent study also found lesion detection rates (adenomas and sessile serrated lesions [SSLs]) among patients aged 45 to 49 were slightly lower but comparable to those among patients aged 50 to 54.53 Additionally, ADR computed from all colonoscopies did not differ significantly from those based solely on screening colonoscopies.54 Based on this growing body of evidence, the latest ACG/ASGE guidelines now recommend a higher ADR threshold of 35% (40% in men and 30% in women) for patients aged ≥45 years undergoing colonoscopy for an expanded set of indications (screening, surveillance, or diagnostic colonoscopy), and a separate ADR target of 50% in colonoscopies following a positive fecal screening test.22
The limitation of ADR–it does not reward the detection of additional adenomas or serrated polyps–has led to interest in alternative detection targets: advanced ADR, adenoma per colonoscopy, and PDR.22 However, advanced ADR is arguably impractical as a quality metrics because endoscopic size measurements are easily corrupted or gamed. In addition, adenoma per colonoscopy can increase costs due to increased pathology charges and PDR can be corrupted by removal of insignificant polyps.55 Conceptually, non-polypoid lesions, such as Paris classification 0–IIc and lateral spreading tumors, could be potential quality assurance targets because these lesions are easily missed and, if incompletely resected, may increase the risk of PCCRC.56 However, incorporating such morphologic measures into quality metrics poses also significant challenges, as they rely on self-reported outcomes that are susceptible to bias, may be manipulated by endoscopists, and difficult to audit objectively.
The rising trend in early-onset CRC is not confined to Western countries; four Asian cohort studies, including data from Korea, revealed an increasing incidence of early-onset in both men and women.57 Moreover, a Korean single-center retrospective study showed a marked increase in adenoma prevalence after the age of 45 years.58 These data underscore the need for tailored adjustments in the age and ADR targets within FIT-based CRC screening programs, addressing the evolving epidemiology of CRC in Korea.

2. Strategies for improving ADR
Colonoscopy is an operator-dependent procedure and scalable interventions have been shown to improve its performance quality (Table 1).59,60 Endoscopist performance significantly impacts future CRC risk not only during screening colonoscopy but even after screening. Along with adenoma characteristics, patients of endoscopists with ADRs below 20% showed markedly higher CRC risks compared to those with ADRs above 20%, with increased risk of 2.35-fold, 2.69-fold, and 2.10-fold following the removal of low-risk adenomas, high-risk adenomas, and after negative colonoscopy, respectively.61 Furthermore, colonoscopists with ADRs just above recommended threshold should continue striving to further improve their ADR, as CRC prevention continues to increase with ADRs exceeding the minimum target. A recent United States study involving 735,396 patients demonstrated that ADRs at or above the median (28.3%) were significantly associated with a lower risk of PCCRC incidence and mortality. Notably, the benefit of lowering PCCRC risk extends into the 40% range of a physician’s ADR.62
Systemic interventions can be implementable at the level of the endoscopy unit to standardize and enhance outcomes. Split-dose bowel preparation is now recommended as the standard preparation strategy.21,34 Compared to regimens taken solely on the day before colonoscopy, split-dose regimens–irrespective of the type and dose of the cleansing agent–were associated with a 26% increase in adenoma detection, a 53% increase in advanced adenoma detection, and a 53% increase in sessile serrated polyp detection.63 To optimize preparation quality, the last dose should be taken within 5 hours of the procedure and completed at least 2 hours before colonoscopy.34 The inclusion of an additional observer can also be beneficial. In a recent meta-analysis of 4 RCTs, participation of nurses during colonoscopy was significantly associated with an increased ADR (OR, 1.19; 95% CI, 1.07 to 1.32).64
Audit and feedback are effective ways for improving the overall performance of endoscopists.65 The use of endoscopist report cards can lead to improvement in ADR by 21%, with low performers deriving a greater benefit of a 62% improvement.66 Similarly, educational interventions targeting independent endoscopists were associated with a significant 29% increase in ADR, particularly a 39% increase in proximal ADR.67 Since the specific content, duration, and modality of educational interventions vary widely, and no single approach has proven superior, it would be beneficial to develop tailored educational interventions, considering the resources available in each endoscopy unit.
Several specific techniques can enhance ADR. There is no doubt that endoscopists should consistently adhere to the fundamental principles of good withdrawal technique, including examining the proximal side of flexures, folds and valves, cleaning and suctioning, adequate distension, and adequate time spent viewing.47 A minimum WT is required but it is not sufficient to achieve adequate inspection. The key difference between endoscopists with the lowest and highest ADR is not WT, but rather withdrawal technique.68 Segmental WT can also aid to improve ADR, with optimal outcomes observed when the WT exceeded 2 minutes, 4 minutes, and 3 minutes for the right-sided, proximal, and left-sided colon, respectively.69 Water exchange, which involves infusion of water during insertion and, in contrast to water immersion, the removal of infused water, debris, bubbles and air pockets during insertion, can achieve the highest ADR and least insertion pain compared with water immersion, air and CO2 insufflation during colonoscopy.70,71 Additionally, as previous tandem studies found higher adenoma miss rates in the right-sided colon,7 a second or retroflexed view of the right side can improve ADR.72 The use of position changes during withdrawal phase enables adequate luminal distension and movement of excess fluid away from the colonic area of interest. A Korean RCT found that dynamic position change–left lateral decubitus for ascending colon and hepatic flexure, supine for transverse colon, and right lateral decubitus for splenic flexure, descending colon, and sigmoid colon–increased ADR (42.4% vs 33.0%) in routine colonoscopy with conscious sedation.73

3. SSL detection rate
Serrated polyps have been increasingly recognized over the last decades as significant premalignant lesions that progress to CRC via the serrated neoplasia pathway, accounting for 15% to 30% of CRCs.74,75 Besides their potential for rapid CRC progression once dysplasia develops, serrated lesions pose significant endoscopic challenges due to their flat morphology, subtle color, and unclear boundaries, both of which are important contributors to PCCRCs.56 The prevalence of SSLs may be rising, or SSL detection may be improving, and recent studies highlight an inverse association between serrated detection indicators and PCCRC. A GIQuIC registry analyzing 2 million screening colonoscopies showed a significant increase in SSL detection rates from 5.0% in 2014 to 7.1% in 2017.76 A Dutch FIT-based CRC screening cohort involving 277,555 colonoscopies reported that each 1% increase in proximal serrated PDR (PSPDR) resulted in a 7% decrease in PCCRC risk.77 Similarly, an Austrian primary colonoscopy screening cohort involving 229,729 colonoscopies found an inverse relationship between PSPDR and PCCRC mortality (HR, 0.97; 95% CI, 0.94 to 0.99).78 The New Hampshire Colonoscopy Registry also demonstrated that endoscopists with higher SSL detection rate (SSLDR) had lower risk of PCCRC.79
Notably, endoscopists with high ADRs may still have low serrated lesion detection rates, with only a moderate correlation between the two.77,78 Given the independent association between serrated detection indicators and PCCRC risk, the latest ACG/ASGE guidelines recommend serrated lesion detection as a separate quality indicator, with a minimum threshold for SSLDR of 6%.22 However, concerns regarding serrated detection targets need to be addressed. First, differentiating SSLs from HPs remains challenging, as a recent study showed only moderate interobserver agreement even among experienced pathologists.80 Furthermore, the 18-fold variations in PSPDR among endoscopists, compared to the 2.8-fold variations in ADR, underscores the need for greater efforts to improve serrated lesion detection.81 Second, several potential serrated detection indicators have been proposed, all aimed at evaluating the detection of relevant serrated lesions (Table 2): PSPDR, SSLDR, serrated PDR, and clinically significant serrated PDR.77-79,82 While the ACG/ASGE guidelines recommend SSLDR as it directly measures the precancerous serrated lesion, its reliability may be hindered by interobserver variation between pathologists. Other indicators have their own limitations; for example, PSPDR excludes large distal serrated polyps, while clinically significant serrated PDR depends on pathologic diagnosis and may be influenced by the subjective measurement of endoscopists regarding polyp size. Third, there are limited data on the prevalence of serrated lesions within FIT-based CRC screening programs. FIT has demonstrated poor sensitivity for detecting serrated lesions; a Taiwan prospective study reported a sensitivity of only 6.2% for SSLs detection, significantly lower than 20.9% for advanced adenomas.83 Similarly, the Evaluation Quality Indicators of the Performance of Endoscopy study, which analyzed 72,021 colonoscopies following a positive FIT, found very low SSLDRs of 1.8%, compared to an ADR of 45%.84 Furthermore, SSL prevalence in studies varies significantly across region, ranging from 2.6% in Asia to 10.5% in Australia.85 Therefore, studies on SSLs prevalence and serrated detection targets suitable for FIT-based CRC screening programs in Korea are needed. In clinical practice, the presence of mucus cap can aid in detecting SSLs during the withdrawal phase.86 Endoscopists should employ strategic washing and be vigilant for SSLs, particularly when adherent stool or debris persistently clings to the mucosa despite cleaning. Additionally, a minimum WT of 6 minutes and the use of chromoendoscopy can enhance serrated lesion detection.87 The New Hampshire Colonoscopy Registry reported an incidence rate ratio of 1.77 for each minute beyond 6 minutes WT, peaking at 9 minutes, similar to the improvement observed for ADR.44 A German prospective study also demonstrated an increase in serrated lesion detection from 29.5% to 46.2% with chromoendoscopy.88 By contrast, image-enhanced endoscopy, including NBI, has shown no clear benefit. A meta-analysis found no significant improvement in overall serrated lesion detection rates (relative risk, 1.16; 95% CI, 0.76 to 1.76)89 and an Australian RCT similarly reported no benefit of NBI for SSLDR (7.5% vs 8.0%).90 To date, no studies have specifically evaluated differences among serrated lesion subtypes, likely due to the relative rarity of SSLs with dysplasia and traditional serrated adenomas.

4. AI in colonoscopy quality
Artificial intelligence (AI), a rapidly advancing technology, is increasingly being applied in the medical field, including in image recognition. In endoscopy, AI algorithms are used for polyp detection and optical diagnosis, commonly referred to as computer-aided detection (CADe) and computer-aided diagnosis.91 As missed lesions remain a leading cause of PCCRC, CADe systems have gained attention as a promising solution to address human error (lesions unrecognized by endoscopists) and improve ADR by serving as a second observer in clinical practice. With the integration of convoluted neural networks and deep learning, real-time CADe systems now assist endoscopists by identifying suspected lesions through visual and auditory alerts. Since Wang et al.92 developed and validated a deep learning algorithm for automatic polyp detection during colonoscopy, multiple RCTs have demonstrated significant improvements in ADR and reductions in missed lesions with CADe. A meta-analysis of initial RCTs confirmed that CADe systems increased ADR compared to colonoscopy without CADe (36.6% vs 25.2%).93 Moreover, CADe has shown enhanced detection of SSLs per colonoscopy.
However, contrary to these promising early results, subsequent real-world studies have not consistently demonstrated the anticipated impact of CADe. A recent meta-analysis of nonrandomized real-world studies of CADe found a statistically significant but clinically minimal improvement in ADR with CADe compared to colonoscopy without CADe (36.3% vs 35.8%; relative risk, 1.13; 95% CI, 1.01 to 1.28).94 Notably, this improvement was only confined to six prospective studies, while six retrospective studies showed no discernible benefit. These conflicting findings underscore the need to refocus on the fundamentals of colonoscopy quality: meticulous mucosal exposure and careful inspection leading to detection. Current CADe systems act as second observers, assuming that the endoscopist adequately visualized the colonic mucosa. However, AI can only analyze what has been visualized; adenomas may still be missed if endoscopists fail to ensure complete mucosal visualization, even with CADe assistance. In this context, recent trials have highlighted the potential benefits of mucosal exposure devices in improving the diagnostic yield of AI-assisted colonoscopy.95-98 For instance, one RCT reported that combining CADe with Endocuff-assisted colonoscopy achieved the highest ADR.95 Similarly, another RCT found that ADR was significantly higher when Endocuff was added to CADe compared to CADe alone (49.6% vs 44.0%).99 To achieve the greatest possible clinical impact, future application of AI support colonoscopy can ensure the completeness of luminal examinations (Fig. 1). Needless to say, clinically relevant lesions–even subtle ones–must not be overlooked to minimize the risk of PCCRC, particularly by nonexperts.100 However, future deep learning algorithms should be designed to assess areas that have been visualized, examined, and potentially missed,101 while simultaneously providing real-time feedback.
The discrepancy between RCTs and real-world studies also suggests that the impact of AI devices is influenced more by various experimental conditions, clinical environments, and equipment used than by the device itself.102 Accordingly, the implementation of new AI technologies should be evaluated based on their actual performance in real-world practice, including their interaction with clinicians.103 Such human-AI interaction can be further optimized through the development of algorithms that minimize false positives and by addressing human factors such as automation bias, alarm fatigue, aversion, and learning effect and deskilling.104 A recently identified challenge is the potential for “de-skilling,” with lower ADR occurring in colonoscopies without CADe once endoscopists had become exposed to, and perhaps somewhat dependent on, CADe.105
Recent advancements in AI technology for the automated evaluation of both detection and process indicators have demonstrated improved performance in colonoscopy. AI-based bowel preparation scales can standardize assessment and even outperform endoscopists’ evaluation by providing a more refined and objective measure during the procedure. A systematic review confirmed the superiority of AI, showing better correlation with expert Boston Bowel Preparation Scale ratings, ADR, and missed lesions.106 In addition, AI can objectively verify cecal intubation, leading to an improved ADR with a notable increase of 5.0%.107 Furthermore, AI-derived monitoring of effective WT–which accounts for both withdrawal speed and the quality of endoscopic images–has shown superiority over standard WT in improving both ADR and SSLDR.108
Given that suboptimal colonoscopy with incomplete mucosal inspection is a key factor for missed lesions, research is increasingly focusing on leveraging AI to assess and improve the overall quality of colonoscopy procedures.109,110 A systematic review described 13 studies encompassing five categories of computer-aided quality assessment systems: withdrawal speed, endoscope movement analysis, effective WT, fold examination quality, and visual haze pattern.110 However, only a limited number of studies have evaluated feedback by implementation the computer-aided quality assessment system and reported their impact on ADR improvement. Further research building on the progress of CADe studies is warranted, with future efforts aimed at assessing the outcomes of integrated approaches to improve colonoscopy quality comprehensively.

PROGRAMMATIC AND FUTURE PERSPECTIVES, AND CONCLUSIONS
High-quality colonoscopy is a key to effective CRC screening, emphasizing the importance of understanding quality indicators and consistently applying them in daily practice (Fig. 2). In FIT-based CRC screening programs, adherence to diagnostic colonoscopy after a positive FIT and timely follow-up colonoscopy are also essential to reducing CRC burden. Failed or delayed follow-up colonoscopy (typically defined as exceeding 6 or 9 months) is associated with an increased risk of CRC.111 Although follow-up targets of at least 80% have been suggested,112 a mixed-methods cohort study from the United States reported a follow-up colonoscopy rate of just 56% within 1 year of a positive stool-based test.113 Identifying barriers to follow-up colonoscopy in FIT-positive populations in Korea and implementing strategies to improve adherence remain critical areas for ongoing efforts.
A Polish colonoscopy screening program demonstrated that a single negative screening colonoscopy predicted significantly reduced CRC incidence and mortality for up to 17.4 years.114 Notably, only high-quality colonoscopy, in itself, was associated with such a profound and stable reduced risk of CRC.
To address variability among endoscopists performing colonoscopy, the Korean Society of Gastrointestinal Endoscopy introduced the National Endoscopy Quality Improvement Program to enhance the quality of endoscopy within the National Cancer Screening Program. The National Endoscopy Quality Improvement Program was revised in 2018 and comprises 29 statements and 34 quality indicators across categories such as the workforce, process, facilities and equipment, outcome, reprocessing, and sedation.20 Compared to current guidelines that propose quality indicators for colonoscopy, these statements–similar to the Asia Pacific Consensus Recommendations on CRC Screening published in 201518–mainly focus on conceptual frameworks rather than actionable implementation (Table 3). The adoption of certain key outcome indicators, such as ADR, has been challenging due to a lack of consensus even among experts. Moreover, optimal performance targets for theses quality indicators were not explicitly defined in the statements. This highlights the necessity for developing country-specific quality metrics and performance targets tailored to the unique healthcare contexts and needs for each region. Ensuring the application of high-quality colonoscopy practices and establishing quality indicators specific to Korea could ultimately lead to improved CRC prevention outcomes.