Endoscopy 2018; 50(10): 950-952
DOI: 10.1055/a-0637-9072
© Georg Thieme Verlag KG Stuttgart · New York

Detection of serrated polyps: How do endoscopists rate?

Referring to Schramm C et al. p. 993–1000 and Crockett S et al. p. 984–992
Joseph C. Anderson
1  Department of Veterans Affairs Medical Center, White River Junction, Vermont, United States
2  The Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, United States
› Author Affiliations
Further Information

Publication History

Publication Date:
27 September 2018 (online)

As serrated polyps are detected in a large proportion of screening examinations [1] and may account for a large proportion of colorectal cancer (CRC), their detection and removal are of major concern to endoscopists. Although most serrated polyps are hyperplastic polyps and have a benign course, a subset of these polyps, known as sessile serrated adenomas/polyps (SSA/Ps), can progress to CRC. Compared with conventional adenomas, serrated polyps, and SSA/Ps in particular, are more likely to be flat and proximally located, suggesting that detection of these lesions may be different. This premise is supported by an analysis demonstrating that the risk for recurrent serrated polyps was much greater than that for conventional adenomas in individuals undergoing endoscopic surveillance [2]. Several studies have also demonstrated great variation in serrated polyp detection among endoscopists [1] [3]. Therefore, a reasonable question is whether a serrated polyp detection rate (SDR) is warranted in addition to the current adenoma detection rate (ADR).

In this issue of Endoscopy, analyses of two separate populations examine the SDR as well as the factors associated with serrated polyp detection. Schramm et al. [4] examined the detection rates for clinically relevant serrated polyps (crSPs), and identified factors associated with crSPs in 4161 average-risk adults who underwent screening colonoscopy at seven academic and community-based practices in Germany. CrSPs were defined as serrated polyps ≥ 10 mm and/or > 5 mm located proximally to the splenic flexure. The investigators observed a large variation in detection of crSPs among endoscopists (0 – 16.2 %). In the study by Crockett et al. [5], the investigators used natural language processing to examine SSA/P detection rates for 104 618 colonoscopies performed by 201 endoscopists at four sites across the USA over a 2-year period (10/2013 – 9/2015). The analysis showed a wide range of SSA/P detection rates (0 – 18.8 %). Thus, both studies demonstrate a wide variation in detection of serrated polyps, supporting a need for a benchmark SDR.

“Proximal serrated polyp detection rate might be a reasonable benchmark to use because of its simplicity.”

However, the studies used different measures. The metric for the US study was histologically based, examining SSA/Ps, whereas the German study used size and location to differentiate the serrated polyps. Which is the best measure to use?

Measuring SSA/P detection as opposed to all serrated polyps is optimal, as these are the lesions that can progress to CRC and may be difficult to detect. In addition, recent data have shown that the presence of SSA/Ps, as opposed to hyperplastic polyps, may have implications for predicting metachronous neoplastic risk [6]. It should be noted that a third type of serrated polyp, known as traditional serrated adenomas, harbor dysplasia and can progress to CRC; however, these lesions are rarely encountered and have a protruding shape that makes them easy to detect and distinguish from other serrated polyps.

However, it can be difficult to distinguish some hyperplastic polyps from SSA/Ps because of similarities in their endoscopic and histologic features. Thus, given the variation and lack of reproducibility in pathologic interpretation of SSA/Ps, this measure may not accurately reflect an endoscopist’s ability to detect serrated polyps. Specifically, credit would only be given to those serrated polyps diagnosed correctly as SSA/Ps. One option may be to measure the detection rate for clinically important or relevant serrated polyps, as was done in the Schramm et al. study as well as previous studies [1]. This definition is based on factors – larger size and proximal location – that may predict SSA/P histology [7], obviating the need to histologically classify the serrated polyps. A drawback is that it might still be cumbersome to use when calculating the SDR. Another option may be to simply use proximal location to classify the serrated polyps (PSDR), as has been done in previous studies [1] [3], obviating the need for classification by size or histology. While this may credit those endoscopists who resect many diminutive proximal serrated polyps, which are likely to be of little clinical significance, the use of proximal location would make the PSDR very simple for endoscopists to calculate. Furthermore, PSDR has been shown to correlate with clinically significant SDR in the study by Schramm et al., as well as in a study conducted using the New Hampshire Colonoscopy Registry (NHCR) [1]. Therefore, PSDR might be a reasonable benchmark to use because of its simplicity. A summary of possible SDR measures is shown in [Table 1].

Table 1

Rates used to measure serrated polyp detection.

Detection rate




Sessile serrated adenoma/polyp (SSA/P)

Serrated polyps diagnosed as SSA/Ps

SSA/Ps have dysplastic potential and are the serrated lesion of concern

Depends on correct serrated polyp pathology interpretation

Clinically significant or relevant serrated polyps

Serrated polyps that are large (≥ 10 mm) or proximal ( ≥ 5 mm) and can sometimes include SSA/Ps

Does not rely on further classification of serrated polyps and includes serrated polyps likely to be SSA/Ps

Can be difficult to calculate, as size and location are needed

Proximal serrated polyps

Serrated polyps located proximally to the sigmoid or splenic flexure

Simple to calculate

Can credit those endoscopists who detect small (< 5 mm) hyperplastic polyps

Another question is what benchmark should be used? The German study also examined PSDRs and observed that the mean detection rate of serrated polyps located proximally to the rectosigmoid or the descending colon were 7.7 % (95 % confidence interval [CI] 5.3 % – 10.1 %) and 6.9 % (95 %CI 5.0 % – 8.9 %), respectively. These rates are similar to those observed previously [1] and meet the 5 % PSDR (proximal to splenic flexure) benchmark rate suggested by one investigator [3].

Do we need a separate quality measure for serrated polyps? One interesting observation in the study by Schramm et al. is that the rate for crSPs was closely linked with ADR. This has also been observed in other studies [1] [3]. However, in the study by Crockett et al., although high ADR correlated with high SSA/P detection rate, there was a significant percentage of endoscopists who achieved an adequate detection rate for one polyp type but not the other. This was also observed in an analysis of the NHCR, where 25 % of endoscopists with a high ADR had a low SDR. Thus, it is probably prudent for an endoscopist to measure their SDR even if they have an adequate ADR.

What steps can be taken by endoscopists to increase their SDR? The US study showed that endoscopists with gastrointestinal training, more years of experience, and higher volume had higher SDRs. The German study observed that an ADR ≥ 25 % was a predictor of a higher SDR, whereas the US study showed that better bowel preparation quality was a predictor of a higher SDR. Thus, both studies suggest that with proper experience and training, as well as adhering to other existing quality measures, endoscopists can achieve an adequate SDR. In addition, another quality measure – withdrawal time – may be helpful. A recent analysis of the NHCR showed that a withdrawal time of at least 8 minutes is associated with higher ADR and SDR [8].

In conclusion, endoscopists should be monitoring their polyp detection rates. If the ADR or SDR are below benchmarks, they should take steps to improve withdrawal technique, increase time for inspection, and optimize bowel preparation. Hopefully, better attention to polyp detection will reduce interval CRC rates.