Authors: Douglas K. Rex (1Division of Gastroenterology and Hepatology, Department of Medicine, Indiana University School of Medicine, Indianapolis, IN), Joseph C. Anderson (2Department of Medicine/Division of Gastroenterology, Geisel School of Medicine at Dartmouth, Hanover, NH; 3Department of Medicine/Division of Gastroenterology, White River Junction VAMC, White River Junction VT; 4University of Connecticut School of Medicine, Farmington CT), Lynn F. Butterly (5Geisel School of Medicine at Dartmouth, Hanover, NH; 6Department of Medicine, Section of Gastroenterology and Hepatology, Dartmouth-Hitchcock Medical Center, Lebanon, NH; 7New Hampshire Colonoscopy Registry, Lebanon, NH), Lukejohn W. Day (8Division of Gastroenterology, Department of Medicine, University of California San Francisco, San Francisco, CA; 9Chief Medical Officer, University of California San Francisco Health System, San Francisco, CA), Jason A. Dominitz (10VA Puget Sound Health Care System, Seattle, WA; 11Division of Gastroenterology, Department of Medicine, University of Washington School of Medicine, Seattle, WA), Tonya Kaltenbach (12Department of Medicine, University of California, San Francisco, CA; 13Division of Gastroenterology, San Francisco Veterans Affairs Medical Center, San Francisco, CA), Uri Ladabaum (14Division of Gastroenterology and Hepatology, Department of Medicine, Stanford University School of Medicine, Stanford, CA), Theodore R. Levin (15Kaiser Permanente Division of Research, Oakland, CA), Aasma Shaukat (16Division of Gastroenterology, Department of Medicine, NYU Grossman School of Medicine, New York, NY; 17New York Harbor Veterans Affairs Health Care System, New York, NY), Jean-Paul Achkar (18Department of Gastroenterology, Hepatology and Nutrition, Digestive Diseases and Surgery Institute, Cleveland Clinic, Cleveland, OH), Francis A. Farraye (19Mayo Clinic, Division of Gastroenterology and Hepatology, 4500 San Pablo Road, Jacksonville, FL), Sunanda V. Kane (20Mayo Clinic Rochester, Rochester, MN), Nicholas J. Shaheen (21Division of Gastroenterology and Hepatology, University of North Carolina, Chapel Hill, NC)
Categories: Article
Source: Gastrointestinal endoscopy
Authors: Douglas K. Rex, Joseph C. Anderson, Lynn F. Butterly, Lukejohn W. Day, Jason A. Dominitz, Tonya Kaltenbach, Uri Ladabaum, Theodore R. Levin, Aasma Shaukat, Jean-Paul Achkar, Francis A. Farraye, Sunanda V. Kane, Nicholas J. Shaheen
Colonoscopy is the cornerstone of colorectal cancer (CRC) prevention worldwide and in the United States^1–4^. In the U.S., colonoscopy is commonly used for primary CRC screening and is the first and preferred colorectal imaging test in patients presenting with symptoms, positive screening tests other than colonoscopy^1–4^, undergoing surveillance after resection of CRC or pre-cancerous polyps^5,6^, strong family history of CRC or advanced precancerous lesions ^1^as well as dysplasia surveillance in ulcerative colitis and Crohn’s colitis^7^.
Evidence indicates colonoscopy reduces the incidence of CRC and prevents CRC mortality^8–23^ (Table 1). Reduction in incidence and mortality of CRC with colonoscopy is greater in the left colon than the right colon^24^. In the first randomized controlled trial comparing colonoscopy to no screening, patients who adhered and underwent colonoscopy (per-protocol analysis) had a 31% reduction in CRC incidence and a 50% reduction in CRC mortality^25^. Several factors, including earlier than planned reporting of trial results, absence of stage shift in CRCs detected in the colonoscopy arm (suggesting symptomatic patients were enrolled in the colonoscopy arm), and lower than expected cecal intubation and adenoma detection, indicate the study may have underestimated the benefits of colonoscopy^25^.
The impact of colonoscopy on CRC and other outcomes (e.g. polyp detection, assignment of screening and surveillance intervals) is highly operator-dependent. Detection of pre-cancerous colorectal lesions is highly variable^26–28^, and is associated with the risk of developing post colonoscopy CRC ^21, 22^. In response to evidence of inconsistent performance, professional gastroenterology and endoscopy societies began an organized movement two decades ago to improve the quality of technical performance and reduce the operator-dependence of colonoscopy ^29^. This document represents the latest update of recommendations from the American Society for Gastrointestinal Endoscopy (ASGE)/ American College of Gastroenterology (ACG) quality task force. Previous recommendations from this task force were published in 2006^30^ and 2015^31^. This update reflects new evidence appearing since 2015.
High quality colonoscopy includes adequate bowel preparation, safe colonoscope insertion to the proximal extent of the colon, detailed examination with identification of all pre-cancerous lesions, and complete and curative resection of these lesions. The process is completed by thorough and accurate documentation of findings, and assignment of any appropriate screening or surveillance follow-up at cost-effective intervals based on recommendations from the U.S. Multi- Society Task Force on CRC^32^. High quality performance in one aspect of colonoscopy does not ensure adequate performance in others. For example, colonoscopists may be effective at detection but not resection of pre-cancerous lesions or vice versa^33^. Understanding deficiencies in performance is generally gained only through quality measurement. Given the impact of inadequate performance on critical outcomes including cancer development, failure to measure performance is unacceptable.
This document presents many quality indicators related to the technical performance of colonoscopy. Practicing colonoscopists are encouraged to make quality measurements related to all the indicators, but this may not be feasible from a time, staffing or cost perspective. Therefore, the document recommends priority quality indicators (Table 2). These indicators were chosen based on clinical relevance, evidence of variable performance, and feasibility of measurement. Measurement of priority indicators is considered essential.
The first version of this document was published by the ASGE/ACG Task Force on Quality in Endoscopy in 2006 ^30 32^ and it was revised in 2015^30, 31^. This current revision integrates new data relevant to existing quality indicators and introduces new indicators based on interval progress in the field. This document focuses on quality indicators unique to colonoscopy (Table 3). The indicators that are common to all GI endoscopic procedures are presented in detail in a separate paper ^34^ and are, for completeness, also listed in Table 4. Indicators common to all GI endoscopic procedures are not addressed in this document, except in some instances where discussion specific to colonoscopy is required.
As in the preceding versions, we prioritized indicators that have wide-ranging clinical implications, are associated with variation in practice and outcomes, and have been validated in clinical studies. When supportive data were absent, indicators of clinical importance were chosen by expert consensus. We have made substantial progress in measuring performance; however, feasibility and efficiency challenges remain. The task force included a limited number of highly relevant, but not yet easily measurable indicators.
As before^30 31^, quality indicators are divided into 3 time pre-procedure, intra-procedure, and post-procedure. Each quality indicator is classified as an outcome or process measure. Outcome measures are impactful in improving quality of care but can be difficult to measure in clinical practice. Outcome measures often require large amounts of data and/or long-term follow-up, and their measurement may be confounded by other factors. In such cases, process indicators are provided as surrogate measures of high-quality endoscopic practice. The relative value of a process indicator hinges on the evidence supporting its association with a clinically relevant outcome, and such process measures are emphasized. The measures in this document pertain to endoscopic care. The quality of care is influenced by additional factors, including those related to endoscopy centers. These structural measures are covered in a separate document dedicated to unit-level quality^35^.
For this update, the task force critically appraised existing quality indicators based on several factors including ongoing relevance and strength of evidence. Additionally, new quality indicators were proposed and, if appropriate, adopted by consensus based on similar considerations. For each indicator, relevant articles were identified by the authors through a systematic search of PubMed (U.S. National Library of Medicine, National Institutes of Health) from January 2014 – the date of the last update of this document – through December 2022. The search strategies for each indicator included a combination of subject headings (MeSH in PubMed) and pertinent keywords. English language restrictions were applied. To identify additional articles, the authors reviewed PubMed’s “similar articles” and manually searched reference lists of relevant papers. Searches were facilitated by health science librarians with expertise in systematic review. Based on this revised literature review, the strength of recommendation for each indicator was evaluated according to a previously employed framework (Table 5). Within this framework, the strength of each quality indicator is divided across a spectrum from “1A”, denoting a strong quality indicator that can be applied to most clinical settings, to “3”, denoting a weak quality indicator due to lack of evidence requiring expert opinion. The strength of recommendation grade for each indicator was established with consensus of the authors.
The process and outcome measures included in this document are attached to a performance target and, therefore, each measure is considered a quality indicator. The task force selected performance targets based on published benchmarking data, informed by literature review. In the absence of available data, when expert consensus considered failure to perform a given quality indicator a “never event,” such as failure to monitor vital signs during sedation, the performance target was expressed as > 98%, since only in exceptional circumstances would the quality indicator not be fulfilled.
Quality indicators are intended to serve as a framework for quality improvement efforts. The included quality indicators and associated performance targets do not necessarily reflect the standard of care, credentialing requirements, or training standards, and it is a misuse to employ any of the quality indicators in this document as such.
The 2006 and 2015 ASGE/ACG quality documents proposed many indicators for quality measurement in the technical performance of colonoscopy^30, 31^. Quality measurement is relatively new and requires resources. The 2015 document proposed priority indicators that every endoscopy unit should endeavor to measure.
The current revision also proposes priority indicators (Table 2). Priority indicators reflect the strength with which they are linked to important colonoscopy outcomes, such as prevention of cancer and cancer death, the feasibility of measurement in clinical practice, evidence for variability in performance, and consensus regarding the significance of the indicator as a continuing or emerging element of high-quality technical performance of colonoscopy. The rationale for selection of priority indicators in this document (Table 2) is largely presented in the discussion of individual indicators below. We recommend that quality improvement efforts initially focus on High Priority Indicators and then progress to other indicators once it is ascertained that endoscopists are performing above recommended thresholds, either at baseline or after corrective interventions.
The cecal intubation rate (CIR) remains a quality indicator for colonoscopy in this update. However, while CIR was a priority indicator in 2015^30^, its status as a priority indicator has been changed in this update (Table 2). High CIRs are associated with prevention of CRC^36^, and failed intubation leads to costs and inconvenience associated with supplementary imaging and repeat attempts at colonoscopy. Thus, we continue to recommend measurement of cecal intubation rates. However, substantial evidence indicates that most gastroenterologists achieve CIRs above recommended thresholds and maintain them^37, 38^. Therefore, sustained measurement of CIR may not be an effective use of resources in units where there is no evidence of inadequate performance.
New priority indicators are noted by an asterisk in Table 2. One of these is adequacy of bowel preparation adequacy which is fundamental to cost-effective colonoscopy. The second is the sessile serrated lesion detection rate (SSLDR), which is selected based on evidence it will contribute to cancer prevention (see below).
Appropriate indications for colonoscopy are listed in Table 6. Each procedure report should include an accepted indication or an explanation if the indication is considered non-standard. Dates and findings from prior colonoscopies (e.g, cancer, advanced lesions, or non-advanced lesions) that drive the current colonoscopy should be documented with the indication.
When colonoscopy is used for appropriate indications there is a higher yield of clinically relevant diagnoses^39 40 41 42^. Accepted indications lists can be used to screen colonoscopy referrals for appropriateness ^43 44 45 46^.
The ASGE/ACG Task Force on Quality indicators for colonoscopy relies heavily on the bowel preparation recommendations of the US Multi-Society Task Force (MSTF)^47^. When bowel preparation is adequate, the subsequent recommended interval for repeat screening or surveillance colonoscopy should be consistent with the MSTF recommendations ^47,6^ unless there is a documented indication for earlier colonoscopy [e.g. an inherited genetic disorder, strong family history, or a disease process such as chronic inflammatory bowel disease (IBD)].
As per the MSTF recommendations, patients with an inadequate preparation should have a repeat study within one year^47^. The MSTF recommended in 2015 that at least 85% of outpatient colonoscopies should be accompanied by an adequate preparation^47^. The European Society of Gastrointestinal Endoscopy (ESGE) recommended a 90% target^48^. The MSTF recommendations are currently under revision and based on continued evidence of high rates of adequate preparation in clinical trials^49^, the ASGE/ACG Task Force recommends the 90% threshold.
Inadequate preparations substantially increase the cost of colonoscopy delivery^50^, and create risk and inconvenience for patients, thus warranting ranking as a priority indicator.
Like the MSTF, we recommend that the colonoscopy report in the case of adequate preparation should include descriptors of bowel preparation as “adequate”, “excellent” or “good”, or record a Boston Bowel Preparation Scale (BBPS) score of ≥ 2 in all three colon segments^51, 52^. Because it utilizes scores and explicit descriptions of what the scores mean, the BBPS is preferred. Further, the BBPS is specifically designed for use after completion of intraprocedural cleaning, which is the most clinically relevant time point to assess cleansing quality. To count toward meeting the 90% adherence target, the recommended follow-up interval must be consistent with MSTF post-polypectomy^32^ or post-cancer resection^53^ surveillance recommendations.
A trained colonoscopist should achieve a high cecal intubation rate (CIR) with a very high level of safety. Cecal intubation is defined as passage of the colonoscope tip proximal to the ileocecal valve, and fully into the cecal caput so that the appendiceal orifice can be identified and photographed, and the medial wall of the cecum between the appendiceal orifice and ileocecal valve can be thoroughly examined^31^. Low CIR has been associated with higher post colonoscopy CRC (PCCRC) rates^36^.
High CIR should be achievable with extremely low rates of insertion related perforation. Mechanical rupture of the colon can be prevented by reduction of loops and bends in the insertion tube, and avoiding pushing against fixed resistance. Use of thinner, more flexible instruments can reduce rupture risk when the colon is diseased from conditions such as severe diverticular disease, radiation injury, chronic dialysis, long-term corticosteroid use, or fixation due to pelvic surgery. Barotrauma perforations occur only in patients with severe sigmoid angulation and narrowing, primarily associated with severe diverticular disease. The risk of barotrauma should be anticipated in all patients with a complex sigmoid colon and the colonoscopist should consider stopping gas insufflation, and using water immersion until the sigmoid is traversed and the proximal colon decompressed^54^.
CIRs above 95% for all indications are readily achievable by high percentages of independently practicing colonoscopists^37, 55–58^. All independently practicing colonoscopists should have CIR measured until it is consistently above the recommended target. After this, continued CIR measurement is optional and can be performed intermittently or not at all in the case of a consistent track record of performance above recommended thresholds. Limited evidence of variable performance for many practicing gastroenterologists, evidence that high performance is sustained, and the obvious ceiling level of 100% justify reduction in measurement for high level performers.
The adenoma detection rate (ADR) is the most clinically relevant and best validated quality indicator in colonoscopy. ADR was proposed in 2002^29^ in response to evidence that adenoma detection and cancer prevention were variable in clinical practice. ADR was proposed in 2002 as the percentage of patients aged 50 years and older undergoing colonoscopy for indications other than IBD or polyposis syndromes who had ≥ 1 conventional adenoma resected and verified by pathology. The initial proposed threshold was 20% (25% in men and 15% in women) and was arbitrarily set at a level below the mean prevalence of adenomas detected at initial screening colonoscopy studies performed in average risk individuals^59–62^. In 2006 the recommendation for ADR was changed to the fraction of patients age 50 years and older having ≥ 1 conventional adenoma in first time screening colonoscopy^30^. This change acknowledged that the original target levels were derived from screening colonoscopy studies^59–62^. In 2010, a large screening colonoscopy study showed that hazard ratios for post colonoscopy CRC (PCCRC) were increased tenfold when colonoscopy was performed by colonoscopists with ADRs < 20% compared to colonoscopy performed by colonoscopists with ADRs > 20%^21^. In 2015, the minimum acceptable thresholds were increased to 25% (30% in men and 20% in women)^31^ after a large study showed further gains in protection against PCCRC with ADRs above 20%^22^. Since 2015, changes in screening guidelines and new evidence support additional modifications to ADR.
First, an increase in early onset CRC (EOCRC) lead to reductions in the recommended age to start CRC screening to 45 years^2–4, 63^. To keep the ADR measurement consistent with the currently recommended age to begin screening, we now recommend that ADR be measured in persons age ≥ 45 years rather than 50 years. Since adenoma prevalence is related primarily to age, this change might warrant a downward adjustment in the minimum acceptable threshold. However, recent studies found adenoma prevalence in 45–49 year-olds to be only slightly lower than in 50–54 year-olds^64–71^ (Table 7). Since mean ADRs have been steadily increasing over the past decade^72^, and the proportion of patients undergoing colonoscopy who are in the 45–49 year old age group is still relatively small compared to persons age ≥ 50 years (Table 6), we recommend no adjustment in the minimum acceptable threshold for ADR is warranted for inclusion of 45–49 year olds in the ADR measurement.
Second, several studies reported the incidence of adenomas at a second screening colonoscopy performed 10 years after an initial negative screening colonoscopy^23, 68, 73–76^ (Table 8). Adenoma incidence at 10 years is just slightly below that identified at initial screening colonoscopy, despite patients being 10 years older. In light of this evidence, we recommend that second and subsequent screening colonoscopies can be included in the ADR calculation without adjustment for the minimum acceptable ADR threshold.
Third, as noted above, the original definition of ADR included surveillance (colonoscopy for prior neoplasia) and diagnostic examinations (colonoscopy for symptoms)^29^. The recommendation to confine the ADR measurement to screening was made in 2006^30^ because the 2002 targets had been based on screening studies. In this update we recommend expanding the definition of routine ADR to again include both surveillance and diagnostic examinations not performed for positive fecal tests. Given the potential for heterogeneity in adenoma prevalence based on colonoscopy indication, one logical concern is that broadening the ADR calculation to include all surveillance and diagnostic colonoscopies might introduce bias based on patient mix. However, in the aggregate, any bias introduced by heterogeneity is expected to be modest. In general, ADRs for post-polypectomy surveillance colonoscopies are approximately 7–12% higher than ADR for screening colonoscopies^22, 67, 77–84^ (Table 9), while ADR for diagnostic colonoscopies that do not include significant numbers of positive fecal tests are lower than screening colonoscopy. Thus, a combined ADR including screening, surveillance and non-fecal test diagnostic colonoscopies is often similar to the screening ADR ^82 85^. A modeling study found that a combined ADR stratified endoscopists into high and low detectors as accurately as the screening ADR over a range of indication distributions^82^.
Notably excluded from the routine ADR calculation are colonoscopies performed for positive fecal testing. While ADR is predictive of PCCRC in patients with positive fecal blood tests^86^, ADR consistently runs at least an absolute 15% higher in patients with positive fecal tests^87–91^. Therefore, positive fecal tests are distinct from other diagnostic indications. The use of fecal testing varies regionally and across health practices in the U.S., and therefore the fraction of colonoscopies performed for positive fecal blood tests is expected to vary substantially across practices. In settings where the proportion of examinations for positive FIT is substantial (as in organized screening programs), and FIT positive evaluations are distributed similarly across endoscopists, FIT positive examinations can be included in the general calculation of ADR for purposes of internal comparisons of endoscopists ^92^. However, inclusion of fecal positive tests in the general ADR calculation makes comparison of ADRs across U.S. institutions and comparison with international results challenging. Therefore, we recommend that ADR in FIT positive patients usually be measured separately (see below). Further, if detection performance is measured in inpatients, it is reasonable to exclude patients from ADR and all other detection measures when the indication would not be reasonably expected to include polyp resection (e.g. colonoscopy for acute major lower GI hemorrhage or colonic decompression).
Finally, multiple reports indicate ADRs have risen steadily over the past decade^70, 93, 94^. In a large registry, mean screening ADR in U.S. gastroenterology practices had risen to 38% by 2018^70^. In a 2014 study, the risk of PCCRC and fatal PCCRC were shown to decrease by 3% and 5%, respectively, for each 1% increase in ADR.^22^. In a subsequent study, this relationship between ADR and PCCRC was shown to hold during the interval 2010–2018, when ADRs were generally higher, and the relationship extended to ADRs above 40%^93^. Further, the relationship holds for colonoscopy performed for non-screening indications ^92,93^. Improvements in ADR in U.S. gastroenterology practices likely reflect the emphasis on quality measurement over the past two decades, plus the widespread availability of high definition colonoscopes.
Given the various considerations outlined above, we recommend that ADR calculations include screening, surveillance and diagnostic colonoscopy, but exclude the indications of positive non-colonoscopy screening test [e.g., fecal blood test, multi-target stool DNA (mt-sDNA), CT colonography], therapeutic procedures for resection or treatment of known neoplasia, cancer genetic (e.g., polyposis) syndromes, and IBD. Specific recommendations for ADR in patients with positive fecal screening tests are discussed below.
Considering that ADR has risen progressively in screening populations^72^, and that inclusion of surveillance procedures may raise ADR by a significant extent for some colonoscopists, we now recommend a minimum threshold of 35% for ADR (40% in men and 30% in women). We acknowledge that the actual level of ADR chosen as the minimum acceptable threshold (and therefore the fraction of endoscopists who are below the chosen threshold) carries an arbitrary element, but the considerations outlined above support this recommendation.
Colonoscopists with ADRs below 35% are recommended to undertake remedial measures to improve and to achieve acceptable performance. Because there is evidence that PCCRC risk continues to fall as ADRs move above 35%, we also recommend that all colonoscopists strive to achieve ADRs well above the minimum recommended threshold of 35%, and indeed ADRs can rise well above 40%^70, 93–97^.
Including more colonoscopies in the ADR measurement has potential advantages and disadvantages. Expanding the included indications results in a larger sample size allowing benchmarking of a higher proportion of the practitioner’s colonoscopies. With a higher number of included colonoscopies, the confidence interval is narrower, and performance conclusions are more accurate, which is particularly useful for low volume endoscopists^98^. The evidence that ADR predicts the risk of PCCRC after non-screening examinations^22, 93 92^ and also in patients with positive fecal blood tests^86^ supports measuring performance across a wider range of colonoscopy indications. Adding more examinations to the ADR measurement however, requires more resources to review and record pathology results, particularly if the process is performed manually. Automatic data extraction eliminates this issue but is frequently not available.
For all measures related to lesion detection, we recommend (if feasible and practical in the context of local measurement programs) excluding examinations where the procedure is aborted during insertion because of inadequate preparation, or the examination is complete to the cecum but is rescheduled within 1 year. In both cases, we recommend photo documentation of poor preparation to support the need for repeat examination. We also recommend exclusion of procedures that are incomplete to the cecum or an ileo-colic anastomosis because of abnormal colonic anatomy (e.g.,marked redundancy or severe sigmoid angulation and narrowing).
We continue to recommend exclusion of colonoscopies for positive fecal screening tests from the routine ADR calculation as discussed above. This exclusion is partly because the fraction of colonoscopies performed for positive fecal screening tests varies widely. The fraction is often low in US centers where opportunistic screening is performed and can be substantially higher in healthcare systems with organized FIT based screening^99^. Thus, and because ADR is higher in these patients, exclusion of positive fecal screening test from routine ADR calculations allows better comparison of ADR across healthcare systems.
There will be instances where the number of colonoscopies performed for positive fecal screening tests is sufficiently high to warrant measurement of ADR within this patient group. ADR in the FIT positive population does predict PCCRC risk^86^. The prevalence of adenomas in FIT positive patients is well documented and can reach as high as 70%^100^. ADR is higher when the cut off level of hemoglobin per gram feces for a positive test is higher^100^. Like the primary screening population, adenoma prevalence in men with positive FIT is typically 10–15% higher than women^100^. In the US, a cut-off of 20 μg hemoglobin per gram feces is often used in FIT assays^100^.
Based on available evidence, we recommend an ADR of 50% (55% in men and 45% in women) in FIT positive populations where assays with cut-off of 20 mcg hemoglobin per gm feces are used. There is less evidence about expected ADR in patients with mt-sDNA test positive ^88, 101–104^. Although evidence is limited, we currently recommend the same threshold for ADR in mts-DNA positive patients as for FIT positive patients.
The ADR has strengths and weaknesses as a quality indicator. Strengths include that it is a total colon measure, that differentiation of conventional adenomas from lesions in the serrated category by pathologists is generally straightforward with limited interobserver variation ^105^, and that ADR is validated as a predictor of PCCRC^106^. A principal weakness of ADR is that it does not reward detection of additional adenomas after the first adenoma has been identified and resected, and many patients have multiple adenomas. This shortcoming has led to measures such as adenomas per colonoscopy (APC), which rewards detection and resection of each adenoma. Like ADR, higher APC has been associated with lower risk of PCCRC^107^.
In general, the correlation between ADR and APC is high^108–111^. Despite this correlation, there are documented instances in which adequate ADR has been associated with low APC in important fractions of endoscopists^110, 112^. For example, in one study 47.6% of endoscopists in the lowest quartile of APC performance had ADR ≥ 25%, while none of the endoscopists in the highest quartile of APC performance had ADR < 25%^110^. Based on these features, we recommend that APC can be considered as an alternative to ADR in practices where ADRs are generally high, and endoscopists are seeking additional information that can identify and discriminate suboptimal performance. In this regard a minimum threshold for APC of 0.6 is recommended^106, 108–111, 113, 114^, but levels above 1.2 are achievable ^96^. APC may not provide additional important information when ADR is low.
The major downside of APC is the potential for a perceived incentive to separate adenomas into separate preservative bottles for pathologic examination. Such a practice would increase costs, and is not recommended to improve the accuracy of tabulating APC. An alternative approach is to record the number of individual adenomas detected and photographed in the procedure report. The photographs serve to support the total number of adenomas identified in the colonoscopy procedure. This approach could be used as an alternative way of counting ADR or APC with a “resect and discard” strategy. In the absence of resect and discard practice, however, photography can still be used to document the number of adenomas identified so that submission of adenomas in separate bottles is minimized^115^.
The ASGE/ACG Quality Task Force on colonoscopy committee considered carefully and at length whether to add a detection target directed toward serrated lesions. A survey found that many US endoscopists are still not measuring ADR^116^. Adding a separate indicator for detection of serrated lesions creates an additional measurement and resource burden for endoscopists and endoscopy units. Further, there are unique challenges associated with implementation of any serrated indicator that are not relevant to ADR. First, there is substantial interobserver variation among pathologists in the diagnosis of sessile serrated lesion (SSL) versus hyperplastic polyps (HPs)^117–120^. This is not an issue in the pathologic diagnosis of conventional adenomas ^31^. Second, a serrated indicator should ideally disincentivize the identification and resection of diminutive distal colon hyperplastic polyps, which are generally considered to not be precancerous^121^. However, an indicator that relies on endoscopist localization during colonoscopy could be subject to bias and gaming in prospective use^106^. Third, some measures of serrated detection have utilized polyp size^122^. These measurements would be subject to substantial endoscopist bias and gaming in prospective use. Despite these cautionary notes, 3 recent studies found that ADR is insufficient as a sole detection indicator predicting PCCRC^122–124^. In two trials, ADR detection above 25% and low serrated lesion detection were found in 13–14% of endoscopists ^123 124^, and another 37–45% of endoscopists had ADR below 25% and low serrated detection. In both studies the correlation between ADR and serrated detection was moderate ^123,124^.
Although a variety of serrated detection indicators have been proposed^106^, the committee considered 3 because of their association with PCCRC in recent studies^122, 123^ (clinically significant serrated polyp, proximal serrated polyp, and SSL detection rates), and we recommend the SSL detection rate (SSLDR) as the quality indicator of choice, because it directly measures the precancerous serrated lesion of greatest interest, and is not subject to endoscopist bias (Table 10).
We acknowledge that SSLDR is subject to measurement and accuracy error related to pathologist interobserver variation in differentiation of SSL from HP^117–120^. Thus, there is a chance that SSLDR will reflect pathologist performance and bias rather than endoscopist performance in some instances^118^. This is a potentially serious flaw in endoscopy detection measures^106^. However, there is evidence that SSL detection is increasing rapidly with time^125^, and that part of this increase reflects improved awareness and education among community pathologists of identification of SSL. Further, introducing SSLDR as a detection target should incentivize endoscopists to work with their pathologists to improve pathologic identification of SSLs^126^. Admittedly, any serrated detection indicator carries inherent flaws for prospective use^106^.
To set the target threshold for minimum SSLDR, we examined prevalence data on SSL as well as the association of SSL detection with PCCRC. A systematic review and meta-analysis found a pooled SSL prevalence of 2.5%^127^. The reported prevalence of SSLs varies between countries and the pooled prevalence in the US was found to be 4.6%^128^. However, there is clear evidence of increasing SSL prevalence over time^125^, with both endoscopist and pathology factors contributing to the increase. A registry study from the US involving more than 5.1 million colonoscopies and 3934 endoscopists found that the prevalence of 4.57% in 2014 increased to 7.14% in 2017^125^. Thus, the prevalence of SSLs as determined by colonoscopic detection is steadily rising.
Data suggest SSLDR may not correlate closely with ADR, and that SSLDR provides predictive data independent from ADR, making it a worthy stand-alone quality indicator. A recent study from the Netherlands performed in FIT positive patients showed that the proximal serrated polyp detection rate (PSPDR) and the SSLDR had similar associations with prevention of PCCRC^123^. In a study from Austria, the magnitude of PCCRC mortality reduction was similar for increases in PSPDR and SSLDR, but the effect on mortality reduction did not reach significance for SSLDR, probably because SSLs are less prevalent than PSPs^124^. Greater prevalence is an advantage for PSPDR as a detection measure compared to SSLDR, but it should be recalled that over half of patients with serrated polyps have them only in the distal colon^124^, which could subject PSPDR to gaming with prospective use. Data from the New Hampshire Colonoscopy Registry also demonstrate that SSL detection predicts PCCRC independent of ADR^129^. The unadjusted risks of PCCRC were 1.4%, 0.6%, 0.6%, 0.4% and 0.3% with SSLDRs of < 1.0 %, 1-<2.0%, 2.0-<4.0%, 4.0-<6.0% and ≥ 6.0%, respectively. Approximately one-third of endoscopists had adequate ADR but SSLDR <6% (33.8%;50/148).
One challenge with SSLDR is that the prevalence of SSLs is lower than that of conventional adenomas, so that substantial procedure volumes will be required for some physicians to determine with confidence if detection is clearly adequate or suboptimal. This same issue impacts the reliability of ADR measurement^98^, but does not negate the value of measurement. It seems reasonable to report SSLDR to colonoscopists, with the expectation the confidence interval around SSLDR will narrow with continued measurement. Assessment and reporting of SSLDR performance over longer time periods than used for ADR may be appropriate.
Based on available evidence, we recommend a current minimum threshold for SSLDR of 6%. There is an expectation this will be revised upward as evidence of increasing detection occurs. Although evidence is limited, we consider that an SSLDR in fecal immunochemical test (FIT) positive patients of 6% (same target recommended for SSLDR in general population measurement) is appropriate, as there is a little evidence that FIT detects serrated lesions, and few data on the yield of serrated lesions in FIT positive patients^130^. Mt-sDNA does detect some serrated lesions based on methylation markers included in the test, but an appropriate target for SSLDR in this population is currently uncertain. One recent study found a 20.5% prevalence of SSLs in mt-sDNA positive patients ^131^.
We recommend that as groups of endoscopists using the same pathology service begin measuring SSLDR they review the range of SSLDRs across the endoscopy group relative to ADRs. If SSLDRs are low across the entire group, particularly in the setting of adequate ADRs, review of SSL diagnosis with group endoscopists and/or local pathologists is needed.
The recommended minimum average withdrawal time in normal colonoscopies without polypectomy, biopsy, or therapy reflects the best evidence available regarding the time most colonoscopists need to apply careful and detailed inspection of the colon, from the appendiceal orifice to completion of retroflexion in the rectum. The first recommendation for withdrawal time made in 2002^29^ recommended an average of 6–10 minutes in normal colonoscopies, based on the observation of two physicians participating in a miss rate study who had mean withdrawal times of 8 minutes and the lowest observed miss rates among 26 endoscopists in the study^132^. In 2006, the recommendation for minimum average withdrawal time was changed to 6 minutes^30^ on the basis of a landmark study showing that in a group of private practice gastroenterologists performing colonoscopy, 6 minutes of withdrawal time in normal colonoscopies was associated with reasonable separation between high and low level detectors^26^. Since 2006, but particularly since 2015, considerable evidence has demonstrated that optimal detection typically requires at least 8–9 minutes average withdrawal time in normal colonoscopies^26, 133–148^ (Table 1). This evidence includes retrospective data showing that prevention of PCCRC is optimized at withdrawal times of at least 8–9 minutes^149^. Observational studies suggest that overall detection of adenomas is optimized at 8–9 minutes or longer^26, 133–148^ (Table 11), as is detection of serrated lesions^136, 139, 143^. A parallel design randomized trial comparing 6 to 9 minutes found 9 minutes to be superior for detection^147^, as did a randomized tandem study comparing adenoma miss rates at 9 minutes versus 6 minutes^148^.
Importantly, withdrawal time is not by itself an adequate measure of detection performance or skill. Rather, the primary measures of detection performance are ADR and SSLDR. Withdrawal time cannot substitute for ADR or SSLDR. Withdrawal time should be measured and recorded in clinical practice, and if ADR or SSLDR are low, an accompanying short withdrawal time suggests inadequate withdrawal technique. Corrective measures designed to improve ADR and SSLDR should focus on optimized examination technique, which consists of systematic attempts to expose the mucosa proximal to each haustral fold, flexure, and valve in the colorectum, achieving adequate distention of the colon, and intraprocedural cleansing to expose mucosal surfaces covered by fluid collections, bubbles or particulate debris^150, 151^. The point is that application of high-quality inspection techniques requires an average of at least 8–9 minutes.
There is some evidence that mucosal exposure devices (specifically Endocuff Vision; Olympus America Corporation, Center Valley, PA) can allow faster examination during withdrawal without compromising detection^152^. Additional study of this concept is needed.
The recommendation that the average minimum withdrawal time should be at least 8–9 minutes in normal colonoscopies should not be translated to mean that best practice or the standard of medical care requires that every withdrawal time should last ≥ 8 minutes. On the other hand, if ADR and/or SSLDR are below minimum recommended threshold and accompanied by an average minimum withdrawal time of < 8 minutes, this could be construed as evidence of inadequate withdrawal technique. Thus, low ADR accompanied by short average withdrawal time should be considered an indication to evaluate and correct inspection technique.
indicator
The U.S. Multi-Society Task Force on Colorectal Cancer (MSTF) ^153^ and the ESGE ^154^ have created detailed recommendations for the resection of colorectal neoplasms. These recommendations cover the assessment and resection of the full range of colorectal neoplasms, including assessment and characterization of neoplasia, selection of resection methods, and optimal methodology for resection. Herein, we focus on quality measurement.
The writing committee considered a range of potential quality indicators for resection and now recommends two indicators that are characterized by feasibility of measurement, evidence of variable practice, and/or agreement among expert endoscopists regarding best practice. Both have substantial clinical relevance.
The first indicator asks that colonoscopists report for all polyps their size, shape, location in the colon by segment (or distance in cm from the anus for left colon polyps), and the method of removal (i.e. cold snare, cold forceps, hot snare, hot forceps, cold endoscopic mucosal resection (EMR), hot EMR, etc.), though we note that hot forceps play no role in colorectal neoplasia resection except for avulsion during hot EMR or the removal of flat polyp overlying fibrosis in the case of a recurrence or a previously partly resected polyp.
Size is an important parameter to record as it drives surveillance. The use of terms such as “small” is discouraged. Terminal digit rounding^155^ is discouraged, and polyps ≥ 10 mm in size should be photographed with a snare of known size fully opened and placed over and against the polyp^153^. Smaller polyps can be photographed using the same technique or placing the tip of the snare sheath (diameter 2.4 mm) up against the polyp base. Polyps in the same colorectal segment can be recorded with a range of sizes and shapes (e.g., 4 flat and sessile polyps 2–4 mm in size removed by cold snare). Polyps ≥ 10 mm should be described individually. If multiple polyps ≥ 10 mm are removed from the same section and by the same method, with similar histologic predictions, they may be grouped together in the report. Certainly, any lesion, irrespective of size, that suggests advanced histology or endoscopic features of cancer (submucosal invasion or deeper) based on morphology or pit pattern analysis should be individually submitted for pathology; and not combined in the same bottle as polyps from different segments.
With regard to polyp shape, the Paris classification is recommended^156^. Use of the terms “sessile”, “flat” and “pedunculated” is also acceptable. From the perspective of management of unexpected malignancy, the most important distinction is between pedunculated and non-pedunculated, since the histologic features that drive the decision for adjuvant surgical therapy vary slightly between these groups^157^. Interobserver agreement in differentiation of Paris IIa from Is lesions is poor, and by extension so is the differentiation of “flat” from “sessile”^158^. Regardless, careful assessment of polyps for these features helps emphasize the extreme subtlety of many lesions, and therefore the need for very careful inspection of the entire colorectum. Lateral spreading lesions, which grow laterally along the colon wall for > 1 cm, can be described by their specific morphology ^153 154^. Accurate assessment of morphology, together with colorectal location (colon vs rectum), can guide decisions regarding piecemeal vs en bloc resection ^153 154^.
Polyp location should be recorded as it facilitates identification of polyp resection scars at surveillance, is essential when malignancy is suspected or reported unexpectedly, and is important to identify serrated polyposis syndrome.
Reporting the resection method is essential. Resection methods, particularly in the use and type of cautery, differ in their complication rates. Resection methods also differ in rates of completeness and recurrence, with forceps associated with higher incomplete resection rates of lesions > 3mm and cold EMR having higher recurrence rates for adenomas ≥ 20 mm in size^159^. Clip placement to treat or prevent hemorrhage is associated with artifacts at follow-up examination^160^. Clip artifacts can be accurately distinguished from residual polyp by high definition imaging^161^, but confirmation of clip artifact is made easier by an accurate initial report of clip placement. Thus, all therapeutic steps of lesion resection should be described in the procedure report including naming all specific devices and whether each resection step employed electrocautery.
The second quality of resection indicator recommended is the percentage of lesions 4–9 mm in size removed by cold snare polypectomy. The first principle guiding the optimal resection of lesions in this size range is the extremely low risk of submucosally invasive cancer, which has decreased over time in successive publications^162^. The advantage of cold over hot resection is the near elimination of perforation and delayed hemorrhage^163–166^, which seems important for a set of lesions that are unlikely to harm patients in the near term. Indeed, the prevalence of adenomas is much higher than the lifetime risk of CRC, so that most patients having adenomas will never develop cancer. This is an important rationale for resecting lesions with minimal risk of harm. Cold snare resection is associated with a more superficial resection plane than hot snare resection^167, 168^, which may underlie the higher risk of residual polyp when cold EMR is used to remove large adenomas^159^. However, randomized trials indicate complete resection rates using cold snaring for lesions < 10 mm are comparable to hot snaring^169–176^. Accurate technique and snare placement are more important than the use of specialty thin wire vs standard snares^177^. The indicator excludes lesions < 4 mm, because evidence indicates that cold forceps and cold snaring are comparably effective for lesions up to 3 mm in size^178–184^, and some practitioners prefer cold forceps. However, cold forceps are less effective than cold snaring for lesions > 3mm in size^178–184^. Thus, cold snaring is the technique of choice for lesions 4–9 mm in size. Pedunculated lesions in this size range can also be removed safely and effectively by cold snaring^185, 186^. Despite this, many endoscopists use cold forceps for lesions > 3 mm in size, indicating there is both substantial evidence of variable practice with regard to resection technique for lesions 4–9 mm, and considerable room for improvement. Many patients have one or more lesions ≤ 3 mm and one or more lesions 4–9 mm in size. Use of cold snaring for the entire range of lesions < 10 mm in size saves the use of cold forceps and reduces plastic waste associated with colonoscopy^187^.
Inadequate lesion resection and variation in resection quality are major issues in colonoscopy quality. The committee encourages developing internal institutional programs that evaluate resection performance utilizing polypectomy assessment tools and provide periodic feedback to endoscopists. Detailed evaluation of PCCRC suggests that a significant minority of PCCRCs result from ineffective resection of advanced lesions^188–190^. Biopsy of the margins of lesions that appeared to be completely resected showed that residual neoplasia may still be present, and that the risk is operator dependent. Efficacy of complete resection varied by 3-fold among operators in one study^191^. Assessment of competency of resection by blinded review of video recordings found that assessed competency also varied 3-fold, with some operators having rates of competent resection as low as 30%, and no correlation between ADR and competency of resection ^33^. The best approach to assess resection competency is through real-time assessment by a monitor, or by expert review of video recordings. Two scales have been validated for this purpose, including the Direct Observation of Polypectomy Skills (DOPyS)^192^ and the simpler and shorter Cold Snare Polypectomy Assessment Tool (CSPAT)^193^ for the assessment of cold snare technique only. The writing committee of the ASGE/ACG Task Force considered that a broad recommendation to systematically evaluate all U.S. endoscopists with these tools would be challenging to implement, since videorecording is not widely available in the U.S., and the practice of real time monitoring would be viewed as costly and impractical by many centers. Nonetheless, we encourage endoscopy centers to invest in recording equipment that can be used in monitoring of both inspection and resection quality.
In its last recommendations, the ASGE/ACG Task Force recommended that quality monitoring evaluate the fraction of non-pedunculated lesions < 2 cm that are referred for surgical resection as a quality indicator^31^. This general topic is still appropriate for review in quality programs, though the method of audit can be challenging, and we are not including this indicator in this update. Endoscopic resection is the preferred method over surgical resection for all benign colorectal neoplasms regardless of size, shape or location with the possible exception of lesions invading the appendiceal orifice for which the entire margin cannot be exposed during colonoscopy^153^. Endoscopic resection of benign neoplasms is associated with lower morbidity and mortality and lower costs than surgical resection^194–198^. If feasible, one approach is to collaborate with the local pathology department to review the pathology of all surgical colorectal resections and determine whether a local issue of inappropriate referral of benign lesions for surgery exists.
indicator)
Optimal performance of colonoscopy means it is both effective in preventing CRC and cost-effective. Optimal results can be achieved through optimized detection and resection of neoplasia during the procedure, followed by continued screening or surveillance using evidence-based intervals. Currently negative screening colonoscopy should be followed by repeat examinations at an interval of 10 years^1^. Shorter intervals for screening are appropriate only if there is an inadequate bowel preparation (in case of inadequate preparation, a repeat colonoscopy with adequate preparation should be performed within 1 year), or if patients have a high risk family history defined as multiple first degree relatives with CRC or advanced lesions or a first degree relative with CRC or an advanced precancerous lesion at age < 60 years, for which a 5 year interval is appropriate^1^. Current recommendations for post-polypectomy surveillance^32^ and post CRC resection^53^ in the U.S. are summarized in documents from the U.S. Multi-Society Task Force. Performing high quality colonoscopy that optimizes benefit, minimizes harms, and maximizes cost-effectiveness, includes an obligation to know and follow these recommendations for screening and surveillance intervals.
Non-adherence with guidelines for screening and surveillance intervals has been a long-standing issue^199–209^. A recent meta-analysis found that 17 to 25.7% of screening colonoscopies are performed more frequently than indicated or without an adequate indication for early repeat^199^. In community practice, there is substantial overuse of surveillance colonoscopy among low-risk subjects and underuse among subjects with high-risk precancerous lesions^199–204^. Modeling indicates that surveillance contributes to reductions in CRC incidence, and that surveillance is cost-effective, including high intensity surveillance in high risk patients^210^. Reasons for nonadherence to polyp surveillance guidelines have been surveyed^205–208^. Rationales reported for overuse include the perception that patients have risk factors not addressed by guidelines, lack of guideline knowledge, a fear of missed neoplasia, and lack of confidence about the strength of underlying research supporting recommendations^205–208^. However, the evidence supporting current recommendations is substantial^32^, and US recommendations include shorter and thus more aggressive intervals than those used elsewhere in the world, which are based on review of the same evidence^211, 212^. Posters summarizing the post-polypectomy recommendations are available for placement in the endoscopy unit^32^. Clinical decision support tools have been shown to improve adherence with guideline recommendations^213, 214^.
Adverse events in colonoscopy span a range of outcomes. For colonoscopy, serious adverse events (SAEs) include perforation, post-polypectomy bleeding, cardiovascular events related to sedation, unplanned hospitalization, and mortality ^215^. SAEs are reported to be under 1% across large integrated healthcare delivery systems ^216^. In a large integrated healthcare organization in Washington State, SAEs (perforation, bleeding, myocardial infarction, stroke, splenic injury) within 30 days after colonoscopy fluctuated between 0.2 (myocardial infarction) and 0.8 (perforation) per 1,000 screening colonoscopies and between 0.4 (myocardial infarction) and 1.1 (bleeding) per 1,000 follow-up colonoscopies ^217^. Most serious non-gastrointestinal post-colonoscopy events are expected based on background event rates, and not attributable to the colonoscopy ^218^.
Endoscopy units must identify, track, and document SAEs related to colonoscopy. While some events can be immediately identified, other events may not be evident until after discharge from the endoscopy unit. Strategies to capture delayed SAEs include using pre-defined billing codes or natural language processing within an electronic health medical record (EMR), in some cases across multiple healthcare systems. A second, and more common modality, is a post-procedure phone call. Several regulatory bodies, such as The Joint Commission require a phone call within 48 hours of a procedure with anesthesia support ^219^; at the same time, there is no guidance on procedures where moderate/deep sedation is employed. While the optimal timing and frequency of these phone calls is not clearly defined and is impacted by many factors such as staffing resources, endoscopy units should achieve a 98% rate of attempting to call patients within 72 hours of their colonoscopy. Finally, SAEs can have devastating consequences for patients, providers, and healthcare teams. Consequently, when they occur, a focused approach should be utilized to review, analyze, and develop interventions to help prevent or minimize such events from occurring in the future. An appropriate and recommended response to SAEs 1) investigation, 2) comprehensive systematic review, 3) convening a group to conduct a root cause analysis (RCA) for identifying contributory factors, 4) develop corrective actions in areas where improvement can be made to prevent future events and 5) implement system improvements or education/remediation^220–223^.
Perforation can occur during or soon after colonoscopy with over half diagnosed by the endoscopist^224^. About 5% of perforations associated with colonoscopy are fatal ^225–227^. Perforation is most common in the cecum and sigmoid colon^228–230^. Common etiologies are barotrauma, direct mechanical injury through endoscope advancement or dilation, and thermal/electrical injury during a therapeutic maneuver^228, 231^. Perforation in screening examinations is lower with a reported incidence of 0.04%^232^; increases to 0.016–0.8% for diagnostic colonoscopies^233^ and 0.02–8% for therapeutic colonoscopies^233^. Patients with diverticulosis^234^ and irritable bowel syndrome (IBD)^235, 236^ as well as those taking corticosteroids are at increased risk of perforation. Non-gastroenterologists performing colonoscopy ^237^ and lower volume endoscopists^238, 239^ are associated with higher perforation rates. Polypectomy^233^ with electrocautery increases the risk of perforation^240^.
Not exerting force with the endoscope against fixed resistance, removing loops, using a flexible instrument (e.g., pediatric colonoscope or upper endoscope) in narrowed sigmoid colons, using cold snare polypectomy instead of cautery in the resection of diminutive polyps^164, 176, 240–242^, insufflation with carbon dioxide^243^, and submucosal injection during EMR^244^ decrease perforation risk. Endoscopists who recognize a perforation during the procedure should attempt closure with through the scope clips^245^ or large clips that are mounted over the end of the endoscope^246^. All perforations should be monitored by the endoscopy unit medical director and reviewed by a quality improvement committee. This approach can lead to changes in processes and opportunities for endoscopist education which can improve practices that reduce future risk.
Bleeding is the most common adverse event of polypectomy^226, 247, 248^. Bleeding can be immediate (during the procedure) or delayed (after release from the endoscopy unit). Risk factors include polyps > 1 cm in size^248–250^, high number of polyps removed^251–253^, proximal colon location^254, 255^, certain comorbidities^253, 256^, and antiplatelet and/or anticoagulant medications^251^. Older observational data on bleeding rates using fixed power output generators found that low-power coagulation current was associated with increased delayed bleeding and less immediate, with cutting or blended current associated with more immediate and less delayed bleeding^257, 258^. A randomized trial of forced coagulation current vs. the alternating cut-coagulation current Endocut (ERBE, Stuttgart, Baden-Wurttemberg, Germany) using microprocessor-controlled generators found similar rates of delayed bleeding and higher rates of immediate bleeding with Endocut ^259^. Endoscopists should follow guidelines for managing anticoagulant/antiplatelet medications prior to and after colonoscopy^260, 261^. Cold snare polypectomy for small polyps (under 9 mm) reduces delayed hemorrhage^164, 176, 241, 242, 262^. Prophylactic clipping of resection sites should be performed when feasible for lesions over 2 cm located proximal to the splenic flexure which were removed using electrocautery (all 3 criteria fulfilled)^263, 264^. Delayed hemorrhages requiring hospitalization, transfusion, repeat endoscopic or radiographic or surgical procedures for control of bleeding warrant review of polypectomy technique, appropriate use prophylactic measures and other related case aspects.
Immediate bleeding should be treated by endoscopic means and rarely require interventional radiology or operative treatment. Delayed bleeding frequently stops spontaneously^265^. In-hospital observation is appropriate if there is ongoing bleeding, or if the patient has comorbidities, resides far away, or has transportation challenges. Repeat colonoscopy is optional in patients in whom bleeding has ceased. Patients continuing to demonstrate overt gastrointestinal bleeding (i.e., hematochezia or melena) require prompt repeat colonoscopy^265^. Treatment can be by clip application or epinephrine injection in combination with multipolar cautery^265, 266^. Repeat post-polypectomy bleeding is rare once it has stopped spontaneously or has been treated with endoscopic therapy.
Most deaths directly attributable to colonoscopy are cardiopulmonary events related to anesthesia/sedation^215^. While all-cause mortality within 30 days of colonoscopy occurs in 0.07% of patients, nearly all deaths are linked to underlying comorbidities such as cardiopulmonary disease, cirrhosis, and/or neurologic diseases^215^. Any death after colonoscopy warrants a detailed investigation and review by the institution’s quality committee to ascertain underlying causes, determine if the death was preventable, and what improvements can be implemented moving forward.
Documenting the extent and severity of endoscopic inflammation in the colon in patients with ulcerative colitis (UC) is essential in the classification of disease severity, prognosis, and risk for future development of colorectal neoplasia^267^. In general, the endoscopic extent of disease correlates with patient reported outcomes, laboratory tests and prognosis. In addition, disease extent and severity are important factors when choosing medical therapy. The report can specify the extent by description of involved segments, or use the Montreal classification to document extent of disease^268^. In the Montreal system, endoscopic disease limited to the rectum is E1, left-sided colitis E2, and any disease proximal to the splenic flexure E3.
Endoscopic disease activity correlates with prognosis and risk of developing colorectal neoplasia. Objective markers of endoscopic disease activity are needed as symptoms often do not correlate with endoscopic findings. In 2021, the International Organization for the Study of Inflammatory Bowel Disease (IOIBD) updated recommendations for treatment targets in adult patients with IBD^269^. Mucosal healing is now an important clinical target associated with long-term clinical remission, avoidance of colectomy, and corticosteroid-free clinical remission^270^. In UC endoscopic assessment can be performed via sigmoidoscopy or colonoscopy.
Several scoring systems are available to grade endoscopic disease activity^271^. The Mayo endoscopic subscore (MES) first reported in 1987^272^ is easily implemented in clinical practice and although not validated, is commonly used and accepted by the FDA. Several mucosal findings comprise the MES including erythema, assessment of vascular pattern, and presence of friability, spontaneous bleeding, erosions, or ulcerations. Scores range between 0 and 3. A modified Mayo endoscopic score (MMES) was introduced in 2015 to quantify the extent of colonic involvement^273^. In the MMES, the colon is divided into 5 segments, and the score for each segment is added to give a modified score, which is multiplied by maximal extent of inflammation and divided by the number of segments with active inflammation. The MMES correlates with clinical, biologic, and histologic activity and allows a better assessment of partial healing but is not validated.
An alternative endoscopic scoring system for UC first developed in 2012 is the Ulcerative Colitis Endoscopic Index of Severity or UCEIS^274, 275^. Three endoscopic findings comprise the 8 point scale and include assessment of the vascular pattern^267, 268^, presence of bleeding^267–270^, erosions/ulcers^267, 268^ scored in the most severely affected part of the colon. Friability is not a component of the UCEIS. The UCEIS is validated, has satisfactory interobserver agreement, is responsive to treatment effects and able to predict medium to long-term outcomes of patients in clinical remission.
The Ulcerative Colitis Colonoscopic Index of Severity (UCCIS) was developed in 2013 and is calculated by evaluating segmental scores using 4 different variables including granularity, vascular pattern, ulceration, and bleeding/friability^276^. Although validated and with high interobserver agreement, obtaining the final score is complex and may not be simple to use in clinical practice.
Other endoscopic scoring tools are under development to assess mucosal healing^277^. Video training in the use of scales is available ^271^. Training programs have demonstrated improved inter-rater agreement seen with expert endoscopists^278^.
An MES of 0 and 1 are both associated with positive patient outcomes in treatment trials. In STRIDE II, the MES or UCEIS were recommended to assess endoscopic activity with endoscopic healing defined as MES of 0 and UCEIS ≤ 1^269^.
Investigators have advocated for both symptomatic and endoscopic assessment of Crohn’s disease (CD) to best quantify response to therapy. Endoscopic assessment of disease activity is an essential aid to clinical practice. While the Crohn’s Disease Endoscopic Index of Severity (CDEIS) and the Simple Endoscopic Score for Crohn’s Disease (SES-CD) have been validated, a minimally important clinical change in score has not been defined for either instrument.
The CDEIS was developed in 1989^279^. Components include nine separate pseudopolyps, healed ulceration, frank erythema, frankly swollen mucosa, aphthoid ulceration, superficial shallow ulceration, deep ulceration, non-ulcerated stenosis and ulcerated stenosis. CDEIS scoring is relatively complex, with total scores ranging from 0–44 and higher scores denoting worse disease activity. High intraclass correlation coefficients for intra-rater and inter-rater reliability were observed for total CDEIS scores during index development, 0.96 and 0.86 respectively. In a second study, the intra-rater correlation was 0.89 and interrater was 0. 71^280^. Construct validity of the two ranges from 0.75 to 0.83. Responsiveness has been measured as correlation between changes in mean and global evaluation of disease with a correlation coefficient (r) of 0.72^281^.
The SES-CD was developed to overcome concerns associated with the complexity of CDEIS. The components of the SES-CD include ulcer size, proportion of ulcerated surface, proportion of the surface area affected by any disease lesion, and stenosis^282^. Each component is graded from 0–3 and a total score is calculated as the sum of the items for all segments of the colon and terminal ileum. The correlation coefficient for intra-rater reliability was 0.91 for CDEIS and 0.98 for the SES-CD ^282^. In a second study, the intra-rater correlation was 0.91^283^. The responsiveness to change was evaluated with repeat colonoscopy at an average of four months after baseline, and changes measured by the two indices were highly correlated with an r of 0.83^284^.
While both scoring systems are widely available for use, the SES-CD is increasingly favored for both eligibility and outcome assessment given its relative simplicity. The BRIDGe (Building Research in Inflammatory Bowel Disease Globally) group used a modified Delphi approach to identify elements of high-quality Inflammatory Bowel Disease endoscopy reporting ^285^. The panel noted that the use of the SES-CD was appropriate in clinical practice although they recognized that the primary advantage was the use of standardized descriptive terminology of endoscopic findings rather than quantitative assessment. The first STRIDE guideline^286^ recommended absence of ulcers as an adequate endpoint in routine clinical care, but the more recent STRIDE-II guideline endorsed SES-CD. Endoscopic healing is reflected by SES-CD < 3 points or SES-CD ulceration subscore of 0 ^269^.
The Rutgeerts score grades early neo-terminal ileal lesions to predict post-surgical outcomes and identify patients for whom therapy escalation may be necessary ^287^. Lesions are graded on a five point scale of increasing severity where a grade of i0 indicates no lesions in the distal ileum, i1 indicates less than five aphthous lesions in the distal ileum, i2 indicates greater than five aphthous lesions in the distal ileum with normal mucosa between lesions or skip areas of larger lesions or lesions confined to the ileocolonic anastomosis, i3 indicates diffuse aphthous ileitis with diffusely inflamed mucosa, and i4 indicates large ulcers with diffuse mucosal inflammation or nodules or stenosis. Endoscopic recurrence is defined as a score of
i2. Currently, assessment of the reliability, responsiveness, validity and feasibility of the Rutgeerts score is required but the scale has proven straightforward for clinical practice.
In summary, we recommend the use of a specific scoring system to assess disease activity for both UC and CD. This is essential in optimizing accurate classification of disease severity and determining prognosis and risk of complications. The most used scores based on clinical experience and ease of use are shown in Table 12.
There is increased risk of CRC in patients with UC and colonic CD^288^. CRC associated with inflammatory bowel disease (IBD) has poorer outcomes compared to sporadic CRC including younger age at onset and worse survival^289,290^.
Routine surveillance colonoscopies are recommended for IBD patients with colonic disease extending beyond the rectum after 8 years of disease to monitor for development of colonic dysplasia or neoplasia^288,291, 292,267^. For patients with IBD and concomitant primary sclerosing cholangitis (PSC) initiation of surveillance is recommended at time of PSC diagnosis^288,291,292,267^. However, data supporting clinical, or mortality benefits of such strategies are limited and, the quality of available evidence is low^293^.
We recommend that the appropriateness of recommended surveillance intervals be measured after colonoscopy in patients with UC or indeterminate colitis with no dysplasia. The basis for assessing appropriate intervals is presented in Table 13. The recommendations assume an examination of the entire colon with adequate bowel preparation. Though there is uncertainty whether dye-spray or electronic chromoendoscopy add benefit when using a high definition colonoscope, we suggest high definition colonoscopy complemented with one modality of chromoendoscopy with targeted biopsy of suspicious areas^288,291, 292,267^.
Table 13 presents a compilation of four clinical guidelines or clinical practice updates to develop a summary guidance^288,291, 292,267^. We considered that the American College of Gastroenterology guideline had a maximum interval of 3 years^267^ while 3 other guidelines/updates allowed up to 5 years between colonoscopies in select situations^288,291,292^.
The recommendations are based on known risk factors for IBD-associated CRC including duration of disease, disease extent, age at onset, degree of inflammation, presence of colonic strictures, concomitant PSC, family history of CRC and male gender^294^. Shorter time intervals are recommended for those with high-risk features while longer time intervals can be recommended for those with limited risk and who have achieved mucosal healing (Table 13).
The primary purpose of measuring colonoscopy quality indicators is to improve patient care through identification of suboptimal performance and implementation of interventions that address deficiencies. Improvement in colonoscopy quality metrics has been documented for bowel preparation quality, polyp detection and resection, and adherence to surveillance recommendation^295–300^. Improvement in ADR is associated with a reduced risk of PCCRC and cancer-related death^301^.
Though a full discussion of approaches to improve colonoscopy quality is beyond the scope of this document, key principles are discussed. Optimal adenoma detection depends upon adequate bowel preparation, withdrawal technique that maximizes mucosal exposure, high -level recognition of even subtle neoplasia and complete resection of precancerous lesions. A critical first step in colonoscopy quality assurance is the audit and feedback of individual physician performance, with most studies demonstrating feedback is associated with quality improvement^302–306^. When individuals have ADRs below the recommended threshold, a variety of interventions have been found effective (Table 14).
Split-dose or same day (preparation entirely the morning of the procedure) bowel preparation increased ADR in randomized-controlled studies^307^. High quality withdrawal technique includes cleaning pools of retained stool, fluid and mucus, exposing hidden mucosa by systematically probing the proximal sides of folds and achieving adequate distention of the entire colon^150, 151^. A second examination of the right colon, with or without retroflexion, improves detection and is recommended^308, 309^. Distal attachment devices help expose mucosa and improve ADR in many randomized trials^310–313^. Other techniques associated with improved ADR include water exchange during colonoscope insertion^314, 315^ and changing patient position during inspection to keep the examined segment non-dependent^316, 317^. Adequate mucosal exposure and inspection requires time, and a low ADR in combination with short average withdrawal times should be viewed as an indicator of ineffective withdrawal technique, and warrants correction.
Optimized mucosal exposure must be accompanied by training in detection of subtle neoplasia, particularly SSLs. Randomized trials have demonstrated the effectiveness of educational interventions to improve ADR^318–322^. The Task Force recommends instruction in the Paris classification^323^ to emphasize the importance of flat and depressed lesions, and review of photographs of flat and depressed conventional adenomas and serrated lesions^324^.
High-definition endoscopes improve ADR^325–327^, and are considered mandatory in current colonoscopy practice. Finally, technical adjuncts to improve ADR can be considered. Electronic chromoendoscopy with second-generation narrow-band imaging NBI (Olympus Corp. Center Valley, PA) resulted in a higher ADR than white-light colonoscopy (odds ratio (OR) 1.28, 95% CI, 1.05–1.56) in a meta-analysis of randomized, controlled studies^328^, though older versions of NBI and electronic chromoendoscopy did not significantly improve ADR^329^. Linked color imaging (Fujifilm Corp, Valhalla, NY) is also associated with detection gains for both adenomas and SSLs^330, 331^. Artificial intelligence systems for detection of neoplasia have been recently approved in the United States with most randomized controlled studies demonstrating significant improvements in ADR^332–334^ and reduction in the adenoma miss rate^335^. Despite the convincing and consistent evidence from randomized trials, assessments of benefit in clinical practice settings have been negative in several instances^336 337 338^. Generalizability to routine practice settings remains to be demonstrated convincingly.
Successful programmatic colonoscopy quality assurance efforts with demonstrated improvements in patient outcomes have been outlined from the UK National Bowel Cancer Screening Proramme^339–342^, the Netherlands^343^, Poland^318^ and Germany^344^, and the US Veterans Administration^345^, and Kaiser Permanente Northern California^346^.
In summary, correction of poor performance begins with measurement of colonoscopy quality to identify specific areas of concern. Quality improvement may depend upon the employment of multifaceted interventions. As a last resort, removal of privileges to perform colonoscopy may be appropriate if satisfactory performance cannot be achieved, as colonoscopy quality is strongly associated with important patient outcomes, such as PCCRC incidence and mortality^21, 22, 93^. These recommendations hold for colonoscopists in all specialties^347^.