Improving Colonoscopy Quality: What’s Working, What Needs Work
Colonoscopy is the leading colorectal cancer screening strategy in the United States, with considerable evidence showing it protects against colorectal cancer, the second overall leading cause of cancer and cancer-related death in the nation.
However, colonoscopy quality and its protective benefits can vary substantially among endoscopists, with up to two-fifths performing below recommended thresholds. Thus, there have been dedicated efforts to identify quality measures for evaluating and guiding endoscopist performance, and to develop strategies for improving procedural quality.
To better understand the game plan for ensuring patients receive consistently high-quality colonoscopies, Healio Gastroenterology interviewed leading experts in the field of gastrointestinal endoscopy to discuss what strategies are currently working, and conversely, which areas still need work. Quality improvement efforts of note include optimizing bowel prep regimens, improving adenoma detection rates (ADRs) and reducing the rate of incomplete polyp resection, which one expert described as alarmingly high.
Bowel Prep is Fundamental
Experts agreed that ensuring patients receive an adequate bowel prep gives the foundation for a high-quality colonoscopy.
Inadequate bowel prep can result in reduced detection of neoplasia and advanced neoplasia, increased adenoma miss rates and uncertain surveillance intervals, Samir Gupta, MD, MSCS, staff physician at the VA San Diego Healthcare System, and associate professor of clinical medicine at University of California San Diego, said during his presentation on colonoscopy quality improvement at Digestive Disease Week.
“With bowel prep, the rationale seems pretty intuitive,” he said at DDW. “You can’t find what you can’t see.”
Routinely assessing bowel prep quality using an appropriate scale is therefore essential to ensuring it is adequate, Gupta said, with “adequate” recently defined by the U.S. Multi-Society Task Force (USMSTF) on Colorectal Cancer as “cleansing that allows a recommendation of a screening or surveillance interval appropriate to the findings of the examination.”
However, the four-point scale often used in clinical practice is somewhat imprecise for determining adequacy by this definition, so the Task Force has recommended that the ability to detect lesions greater than 5 mm in size is a clinically relevant determinant of bowel prep adequacy.
In terms of validated scales for use in clinical practice, Douglas K. Rex, MD, from Indiana University, who co-authored the USMSTF document, said the Boston Bowel Preparation scale is currently the best option.
“The Boston scale really pertains entirely to what the colon looks like after you’ve cleaned it up, and of course that’s very appropriate for clinical practice,” Rex told Healio Gastroenterology. “If the score is two or higher in all three segments, [it] makes for a very adequate preparation. This score can be used in clinical practice to determine adequacy of preparation, and a score of two or higher in each segment means the screening or surveillance interval appropriate for the findings can be followed.”
A systematic review of bowel prep scales by Parmar and colleagues showed the Boston scale was the most user-friendly and validated scale available, and should therefore be the first choice in a clinical setting. They found increasing Boston scale scores were associated with greater polyp detection in the left (OR = 2.58; 95% CI, 1.34-4.98) and right colon (OR = 1.6; 95% CI, 1.01-2.55), fewer repeat colonoscopies (cutoff of 5, P < .001), and shorter insertion/withdrawal times (P < .001). They also observed substantial-to-excellent interobserver and intraobserver reliability with the Boston scale (intraclass correlation coefficient = 0.74-0.91).
Importantly, a 1-year surveillance interval is now the “clear recommendation” of the Task Force in the event of an inadequate bowel prep, which Rex stands by. “This recommendation puts an end to the process of considering how bad the preparation is in combination with the findings of the exam and then selecting an interval,”he said. “This process of selecting intervals when there is a suboptimal prep is vague and subject to substantial interobserver variation. The new process is much simpler: if the prep is inadequate, the exam should be repeated within a year.”
To minimize the rate of inadequate bowel prep and improve patient acceptance, experts agreed that split-dosing for morning procedures or same-day dosing for afternoon procedures should now be considered the standard of care, as mounting evidence shows these are superior to traditional day-before dosing strategies.
Split-dosing: Standard of Care
A meta-analysis by Martel and colleagues published in 2015 showed split-dosing resulted in significantly better colon cleansing (OR = 2.51; 95% CI, 1.86-3.39) and patients’ willingness to repeat the preparation (OR = 1.9; 95% CI, 1.05-3.46) compared with day-before dosing. Studies have also shown that split-dosing improves adenoma detection rates, leading a number of physician societies to formally endorse split-dosing, the first of which was the 2009 ACG Guidelines on Colorectal Cancer Screening.
The single most important change endoscopists can make to improve the efficacy of their bowel preparation is to begin split-dosing if they are not already doing so, according to Rex.
“If you’re not split dosing, there’s nothing that you can do in the way of changing to a different preparation that will have anywhere near the same magnitude on the quality of prep that you’ll get from going from what you’re currently using in a non-split dose to a split dose,” he said.
The good news is that most endoscopists now appear to be split-dosing, at least based on non-scientific polling. At Gupta’s DDW presentation, for example, the vast majority of the audience raised their hands when asked if they were split-dosing. Still, Gupta said he remains uncertain of how many hold outs there are.
“That was an audience interested in colonoscopy quality,” he pointed out in an interview with Healio Gastroenterology. “I really don’t have a good sense, at this point, of how many practices have not yet moved to using split prep.”
The main barrier to split-dosing is the worry of inconveniencing patients, Gupta explained. A morning procedure, for example, would require the patient to wake up before 4 am to drink the second part of their prep, “but I personally think it’s more inconvenient for the patients who have to repeat the procedure,” he said.
Another barrier is the perception that a patient is more likely to experience incontinence en route to the endoscopy center, but this is largely inaccurate, according to Rex.
“Multiple studies have looked at this, and they do show a numerical increase in the risk of incontinence, but the absolute risk of incontinence is still very low, and it looks like it’s a very low price to pay for the advantages of split dosing,” he said.
Yet another now resolved issue of concern is the possibility of an increased risk for aspiration with a split dose prep, “because of the perceived high volume of fluid in the stomach at the time [patients are] sedated,” according to Michael B. Wallace, MD, from the Mayo Clinic in Jacksonville, Fla.
“Now a series of papers, which we just published in Gastrointestinal Endoscopy, have shown that ... gastric residual volume [is] lower with a split prep than if you drank all the prep the night before,” he told Healio Gastroenterology. “Interestingly, drinking the prep actually facilitates gastric emptying and leaves the stomach more empty, so that has really put that issue to rest. If the anesthesiologist is concerned about drinking a liter or two of prep the morning of the procedure, I think now we can very confidently reassure them that that’s not an issue, certainly if they’re within the normal guidelines of nothing by mouth for the 2 hours immediately prior to sedation.”
Finally, there is no evidence that safety — with respect to electrolyte changes — differs with split dosing vs. traditional dosing, according to Rex, “but my guess, in the long run, is that there will be some slight advantages in regard to safety, because when you split the doses by a number of hours you give the patient time to recover from any electrolyte abnormalities that have been induced by the first dose before they take the second dose.”
While the consensus is that 4 liters of polyethylene glycol electrolyte lavage solution (PEG ELS) given in split doses is the gold standard, Rex said that clinical trials are needed to identify an optimal bowel prep for patients at high risk for bowel prep failure and that more should be done to identify such patients.
There is a tendency to use one favored bowel prep to make the logistics of assigning preps easier and reduce costs for the unit, he said. However, known predictors of inadequate prep include previously poor preparation, chronic constipation, use of opioids and tricyclics, previous bowel resection, diabetes and obesity.
“All of these patients are more likely to require extra preparation, and so it’s important to identify them in the triage process and then use a more aggressive preparation,” he said. On the other hand, a lower volume prep is appropriate for patients who are at low risk of failure, he added.
Importantly, the USMSTF encourages endoscopists to submit procedure reports to data registries for benchmarking performance and quality measures against national thresholds and their peers, and to undergo an improvement initiative if their adequate bowel prep rate is below the target benchmark of 85%.
Data on more than 3 million screening colonoscopies in the nation’s largest GI registry, The GI Quality Improvement Consortium (GIQuIC), show that bowel prep is adequate just under 90% of the time, according to GIQuIC president and director Irving Pike, MD, of the John Muir Health System in Walnut Creek, Calif. “Obviously there are gaps,” he told Healio Gastroenterology.
Adequate bowel prep rates are just one of the 11 colonoscopy quality measures benchmarked by the GIQuIC registry, including ADR, cecal intubation rates, surveillance intervals and withdrawal times, among others. However, ADR is widely considered to be the primary quality indicator for endoscopists, and ongoing efforts by GIQuIC and others aim to track and improve ADRs.
Wide Variation in ADR
ADR, defined as the proportion of screening colonoscopies that detect at least one adenoma, appeared in multiple studies linked to the risk for interval cancer after colonoscopy.
Corley and colleagues’ evaluation of more than 300,000 colonoscopies, published in the New England Journal of Medicine in 2014, showed each 1% increase in ADR was associated with a 3% decrease in risk for cancer (HR = 0.97; 95% CI, 0.96-0.98).
Moreover, ADR varied significantly between colonoscopists. In the Corley study, ADRs ranged from 7.4% to 52.5% among 136 gastroenterologists.
“It varies so much because of skill level in detecting polyps,” Rex said, “and it’s clearly an endoscopist factor. The idea that it has to do with differences between practices, average [patient] age, obesity, diabetes or size is completely incorrect.”
Wallace agreed, citing ongoing research exploring the reasons behind variations in ADR.
“Some physicians explain a low ADR by saying they have particularly low-risk populations, maybe younger patients or more female patients,” he said. “We’ve now looked at a study using the GIQuIC dataset to see if differences within [patient populations] explain these wide variations in ADR, and the bottom line is, they don’t. Even if you have a slightly higher proportion of low-risk patients, that difference is still very small and not sufficient to explain why someone would have a low ADR. So there really is no excuse for having a low ADR.”
The target 25% ADR benchmark recommended by the leading GI societies (20% for women, 30% for men) is therefore not only reasonable but may even be considered low, Wallace said.
“There are large studies, particularly [the Corley study], showing that achieving ADRs even above 33% continue to provide incremental protection against colorectal cancer,” he said. “So the bottom line is, the higher the better. We don’t actually know yet what is the best ADR, and the Corley paper suggests it should be even higher than 25%.”
Pike agreed that endoscopists should strive to exceed a 25% ADR. “We’re finding that many physicians in GIQuIC have ADRs at 40% or above, so colonoscopists should not be comfortable with the low end of the target,” he said.
Furthermore, ongoing research comparing GIQuIC data with performance results from the Corley paper shows that, among more than 4,000 gastroenterologists submitting data to the registry, most are exceeding the target benchmark, Pike said. Initial results presented at the 2015 ACG Annual Meeting showed “approximately 50% of the gastroenterologists in GIQuIC are performing at Corley’s fourth and fifth quintiles, and approximately 85% are performing at the third quintile, which includes the 25% target or higher.”
Rex echoed this sentiment, saying that “the thresholds that have been set for ADR can be easily met by a careful endoscopist who does basically two things: number one, they know the full range of endoscopic appearances of lesions, and number two, they examine a clean colon carefully.”
Nevertheless, as many as one-fifth to two-fifths of endoscopists have below-average ADRs in clinical practice, Gupta said.
On a positive note, ADR is “a solid metric that really correlates with outcomes that we care about,”whereas many quality areas do not yet have a reliable metric, he said. “Where we’re still falling short — and it’s difficult to quantify — is I’m not sure that people are consistently measuring their ADR.”
The main barrier to calculating ADR in the first place is simply time and resources, Gupta said.
“In most practices, a person has to manually review charts to estimate the ADR, and if it’s not being done by manual chart review, you still have to task someone to work with the EHR data,” he said. “To get a reasonable estimate for each colonoscopist, you probably need to review 100 male and 100 female charts,”which is less work than some may think.”
Despite the time investment, experts agreed that it is essential for colonoscopists to know their ADR to know where they stand. Participating in a registry like GIQuIC is one possible solution to streamline this process, according to Pike.
“Collecting data in this way is relatively inexpensive compared to doing so manually, [considering] the cost of a quality assurance staff member, nurse or others collecting the data and abstracting it from reports,” he said. “It’s also a qualified clinical data registry, thus you can submit your data from GIQuIC to CMS and receive credit for your PCRS quality metrics, which will also qualify you moving forward for some of the new quality programs that CMS is rolling out in the coming years.”
Wallace agreed that consistently measuring ADR is essential to improving it, as results from his group’s randomized controlled trial of an endoscopic quality improvement program (EQUIP) suggest a Hawthorne effect, whereby just the knowledge of having their performance measured appeared to improve endoscopists’ ADRs.
“Like almost all things in life, if you know you’re being watched, you’re a little more careful at what you do, and that’s true for colonoscopy as well,” Wallace said. “We know that when people are aware they’re being monitored, they do more careful colonoscopy, and their ADRs go up. We also know that specific programs like feedback on quality, particularly if it’s very focused on things like adenoma detection, have generally been shown to increase ADR.”
The first EQUIP trial, which randomly assigned a group of 15 endoscopists to the educational intervention with or without active feedback, showed a significant increase in endoscopist-specific ADRs (P = .0013) with active feedback. In the larger EQUIP-3 trial involving a total of 22,316 colonoscopies, Wallace and colleagues randomly assigned five out of nine centers to receive the educational intervention with active feedback.
“All of the sites were signed up for GIQuIC, so they knew they were being monitored, but feedback was only given to the intervention sites,” Wallace said.
ADRs increased from 31% to 42% after training at the intervention centers (P = .004), but also increased from 36% to 39% at the control centers, resulting in a nonsignificant difference between groups.
“In our view, this supports the notion that just being aware you’re being monitored, even without much feedback, seems to increase ADR,” Wallace said. Additional measures like active feedback and other educational interventions “seem to have a stronger effect, although the difference in our study was not significant.”
Taken together, the EQUIP studies show this educational intervention works, and Wallace’s group is currently in talks to integrate it with GIQuIC materials. In addition, the EQUIP-3 study is now in press in Gastrointestinal Endoscopy, where these educational materials are accessible. A number of groups, including the U.S. Veterans Administration Hospital and the New York State Department of Health, have already used the EQUIP module as a quality improvement program, Wallace said.
In addition to the EQUIP program, physicians with low ADRs have a lot of options to improve, “which was hard to say about 5 years ago,” Rex said. “These include educational measures to improve their knowledge of the range of subtlety in the endoscopic appearance of precancerous lesions in the colon, education about how to perform withdrawal carefully and devices that can improve mucosal exposure.”
Then again, experts agreed that endoscopists should not underestimate the power of simply examining the colon carefully and thoroughly. One frequently cited example is ensuring procedures are performed with a 6-minute minimum withdrawal time, which was shown to increase ADRs in a preliminary study performed by researchers from Rockford Gastroenterology Associates in Illinois, published in the New England Journal of Medicine in 2006.
According to Pike, these researchers had a journal club within their own group, in which “the two top performers explained to the other endoscopists what it was they did during a colonoscopy, and this, along with using a timer to make sure they were taking their time examining the colon resulted in an improved performance of almost every physician within the group, including the top performers.”
Similarly, data presented by Kushnir and colleagues at the 2014 ACG Annual Meeting showed a second examination of the right colon in either retroflexed or forward view can improve ADRs.
Although promising, technological innovations to improve ADRs — like high definition and wide-angle scopes, and accessories that improve visualization, such as Endocuff, Endorings and Fuse — can be cost prohibitive, according to Wallace.
“The disadvantage of devices is that they’re expensive, and in most cases single use devices, so in colonoscopy where there’s a huge emphasis on cost containment, adding a single use disposable device that has some cost to it is challenging,” he said. “Our focus has really been on the education, and just improved technique ... because that essentially is zero cost other than the short amount of time involved in the education.”
While ADR is an established colonoscopy quality measure, less is known about the quality of endoscopic resection of detected adenomas. Experts agreed, however, that recent data on this quality measure are surprising, if not alarming.
Improving Incomplete Polyp Resection
The widely cited CARE study by Pohl and colleagues demonstrated that the rate of incompletely resected neoplastic polyps in clinical practice is high, which is worrisome considering incomplete resection may contribute to the development of interval cancers.
Of 346 neoplastic polyps removed by 11 gastroenterologists in this study, 10.1% were incompletely resected, and the complete polyp resection rate varied widely among endoscopists, from 6.5% to 22.7%.
“To me, this shows alarmingly high rates of incomplete polyp resection,” Gupta said.
Moreover, this study showed “the rate of incomplete resection gets worse as polyps get larger, and in particular, sessile serrated adenomas have a very high rate of incomplete resection,” Wallace said. “Nearly a third of sessile serrated polyps in their trial were incompletely resected, and that’s probably a low estimate because ... the physicians knew they were being watched.”
Thus, like ADR and bowel prep, measuring the rate of incomplete polyp resection is essential to improve it. However, the best metric to ensure a complete polypectomy is uncertain, according to Gupta.
Taking a careful photo is one possibility, he said. “If multiple colonoscopists looked at the same picture of a polyp removal site, could they consistently agree on when the polyp has been removed or not? That’s a study that’s probably worth doing.”
Another alternative is taking a biopsy of the edge of the polyp removal site, but this could be tedious and difficult to implement in clinical practice, and could be associated with significant bleeding, according to at least one study, Gupta said.
Unfortunately, available data on this quality measure are limited, and innovations on ways to improve complete polyp removal are needed, he said. Drawing on the available literature, he compiled some suggestions about optimizing polyp removal during his DDW presentation, in the form of a mnemonic device called MIINDful polypectomy:
- Move the polyp to the 6 o’clock position;
- Inspect the edges carefully, and consider NBI or dye spray;
- If flat, inject a contrast agent mixed with saline for a submucosal lift;
- Note whether borders are completely removed; and
- Due diligence with photo documentation.
Although none of these strategies have been rigorously shown to improve complete polyp removal rates, these can be considered “rules of thumb, ... but we need to do more research in this area,” Gupta said.
According to Wallace, completely visualizing the polyp is essential for ensuring its complete removal.
“This is particularly true for flat polyps and sessile serrated polyps that are already difficult to see,” he said. “So using HD cameras, optical enhancements like [narrow band imaging] or similar technologies, may improve that.”
For smaller polyps, a growing body of evidence shows that snare-based technique may be more effective than forceps, and that electrocautery may increase complications without increasing complete polyp resection rates.
Therefore, moving toward cold snare techniques for small polyps, and also using saline assisted polypectomy or EMR techniques for large polyps, is probably the most valuable advice based on currently available literature, Wallace said.
Ongoing Research, Education
Experts agreed there is much work still to be done to better understand ways to improve adequate bowel prep rates, ADRs, complete polyp resection and other quality indicators.
On a brighter note, U.S. gastroenterologists have had extremely high insertion rates for a very long time, according to Rex.
“There are lots of practicing gastroenterologists in the U.S. with 98% to 99% insertion rates,” he said. “While I think insertion is an important thing to measure, I don’t think it’s an issue for most American gastroenterologists. Some of the other issues like detection skills and polypectomy skills obviously remain very large problems.”
Fortunately, there are more data from polypectomy trials than ever before, and important research on polypectomy is ongoing, he said.
In the meantime, endoscopists should take advantage of the numerous educational resources available from the ASGE and other societies, Wallace said.
“For the physician who’s already in practice who wants to improve their technique, there are many resources available,” he said. “First and foremost, they should start measuring and get a sense of where they [stand] through programs like GIQuIC, and if they find that their numbers are low, there are many courses available through the ASGE, many lecture sessions and many online resources they can use to improve their detection. But, simply being aware of it is the most important factor.” – by Adam Leitenberger
- Barclay RL, et al. New Engl J Med. 2006;doi:10.1056/NEJMoa055498.
- Corley DA, et al. New Engl J Med. 2014;doi:10.1056/NEJMoa1309086.
- Gupta S. Sp336. Presented at: Digestive Disease Week; May 21-24, 2016; San Diego, CA.
- Johnson DA, et al. Am J Gastroenterol. 2014;doi:10.1038/ajg.2014.272.
- Martel M, et al. Gastroenterol. 2015;doi:10.1053/j.gastro.2015.04.004.
- Pohl H, et al. Gastroenterol. 2013;doi:10.1053/j.gastro.2012.09.043.
- Wallace MB, et al. Gastrointest Endosc. 2016;doi:10.1016/j.gie.2016.07.042.
- Kushnir VM, et al. Abstract #8. Presented at: ACG Annual Scientific Meeting; Oct. 20-22, 2014; Philadelphia, PA.
- For more information:
- Samir Gupta, MD, MSCS, can be reached at email@example.com.
- Irving Pike, MD, can be reached at irving.pike_MD@johnmuirhealth.com.
- Douglas K. Rex, MD, can be reached at firstname.lastname@example.org.
- Michael B. Wallace, MD, can be reached at email@example.com.
Disclosures: Gupta reports consulting for Exact Sciences, Clinical Genomics and Guidepoint Global. Pike is the president and director of GIQuIC. Rex reports honoraria from Boston Scientific; consulting or advisory roles for Boston Scientific, Covidien, Endo-Aid, Ironwood Pharmaceuticals, Novo Nordisk, Olympus and Paion; and research funding from Boston Scientific and Olympus. Wallace reports consulting for Olympus and ownership interests in iLumen.