Variable Endoscopist performance in proximal and distal adenoma detection during colonoscopy: a retrospective cohort study

Colonoscopy is considered the most effective method for the identification and removal of adenomatous polyps [1]. However, multiple studies have shown significant polyp and adenoma miss rates for colonoscopies [2], as well as lower than expected reduction in mortality for proximal colorectal cancers [3]. These factors have highlighted the necessity for effective quality indicators to monitor and advance colon cancer screening programs.

The quality indicator currently recommended by the American Society for Gastrointestinal Endoscopy is Adenoma Detection Rate (ADR) [4, 5]. ADR is defined as the proportion of colonoscopies where at least one adenoma is found [6]. An endoscopist’s ADR has been shown to be associated with overall patient risk of interval colorectal cancer (colorectal cancer diagnosed within a few years of colonoscopy) [7, 8], risk of distal interval colorectal cancer [9], and risk of fatal interval colorectal cancer [8].

Little is known about the rate of adenoma detection in specific segments of the colon. Lesions in the distal and proximal colon have been shown to adhere to differing patterns of development. Specifically, sessile serrated adenomas are primarily observed in the proximal colon. These lesions are more difficult to detect and resect and have been proposed to follow a unique serrated pathway that progresses to carcinogenesis more rapidly than the APC-linked pathway of other adenomas [10‐12]. Whether this variablity in adenoma presentation in different segments of the colon is reflected in variable rates of adenoma detection between segments, remains to be determined. Recent publications and guidelines have focused on identifying adenomas in the proximal colon. To our knowledge, only two studies to date have examined adenoma detection rates in the distal colon [13, 14]. These have produced inconsistent results. Whereas Boroff and colleagues observed a lower inter-operator (between endoscopists) ADR for the distal colon, as compared to the proximal colon [13], Schramm and colleagues observed higher distal ADR [14]. Intra-operator (for the same endoscopist) differences for distal and proximal detection have not been previously assessed.

Identifying variability in an endoscopist’s detection of distal vs proximal adenomas offers an opportunity to enhance performance assessment and improve colonoscopy quality. The aims of the study presented here are to 1) determine if intra-operator differences exist in distal and proximal ADR and 2) identify factors that predict overall, distal, and proximal adenoma detection.

A total of 1907 patients underwent screening colonoscopies and 11 endoscopists were included in the study (Table 1). The indication for the majority of patients was having a first degree relative with colon cancer (63%). The mean age was 60 years old and 42% of patients were male. In most cases (83%) the endoscopist was a gastroenterologist. Very few of the procedures involved poor bowel preparation (1%). The cecum was intubated in almost all of the cases (97%). More than one colonoscopy was performed within 12 months of the index colonoscopy in 70 patients (4%). Patient demographics and procedure details are presented in Table 1.

In this study, we demonstrate a notable variability in endoscopist distal and proximal adenoma detection. Our findings suggest that distal and proximal ADR, used in conjunction with overall ADR, can improve the assessment of colonoscopy quality.

The observed variability in adenoma detection rates between the distal and proximal colon may be attributable to both physiological differences between colon segments and the varibility in operator skill sets. For instance, there are reported differences in segment-specific adenoma occurrence rate [13, 17]. Boroff and colleagues found that while only 30% of distal polyps are diagnosed as adenomas on histology, adenomas correspond to 76% of polyps in the proximal colon [13]. Consistent with these findings, we observed markedly lower distal ADPRQ, as compared to proximal APDRQ (38 and 63%, respectively).

Adenoma detection is also dependent on skill, training and experience, as demonstrated by significant variability in ADR between endoscopists [18]. Interestingly, adenoma miss rates between endoscopists have been previously found to differ for the distal and proximal colon, suggesting that different skills may be required for these segments [12]. In accordance with this observation, our study demonstrates that proficiency in distal adenoma detection does not directly correlate with proximal adenoma detection for the same endoscopist. This observation was independent of endoscopist overall performance (as indicated by overall ADR). Section-specific operator skills necessary for adenoma detection may relate to variances in polyp morphology [19], frequency [10], and the extent of blind spots in certain parts of the distal and proximal colon [20].

Some specific operator skill measures found to be associated with overall ADR include endoscopist accreditation and colonoscopy volume [21], as well as years of experience [22]. Here we confirm that overall ADR is correlated with all of these operator characteristics. However, while proximal adenoma detection was associated with endoscopist procedure volume, that was not the case for distal ADR. This indicates that the increase in ADR among higher volume endoscopists may be related to their ability to detect adenomas in the proximal colon. Endoscopists with a higher degree of skill and experience are more likely to complete the colonoscopy to the cecum and are better attuned to detecting sessile serrated adenomas [23]. In the setting of limited endoscopy time, if more time is attributed to identifying and removing polyps the proximal colon, the withdrawal time for the distal colon may be compromised, resulting in reduced distal adenoma detection [24].

Adenoma detection also showed a weak inverse association with endoscopy years of practice. This is consistent with previous research demonstrating that optimal ADR occurs with a lower number of years of experience [15], suggesting that training plays a more important role in the quality of adenoma detection than years of experience [25].

Patient-related overall ADR determinants are known to include patient age and gender [26]. Our findings confirm these observations and demonstrate similar associations for both distal and proximal ADR.

Certain study limitations should be noted. First, the study was carried out in a single institution setting. Future investigations should determine if our observations are applicable elsewhere. On the other hand, our observed variability in proximal and distal adenoma detection is consistent with what has been previously observed [14]. Second, we were unable to examine certain procedural factors that have been previously associated with ADR such as withdrawal time [27] and time of day [28, 29]. The influence of colonoscopy queue position during the day, insertion time, and withdrawal time on proximal and distal adenoma detection warrant further study. Finally, we did not differentiate between sessile and non-sessile adenomas. Sessile polyps have been shown to be more difficult to detect and may represent higher risk lesions [30, 31]. However, our approach is consistent with the existing endoscopy quality literature, which includes sessile serrated polyps as adenomas [15]. Finally, the overall ADR for this study was 25% which, although within the acceptable quality standards for ADR recommended by the American Society for Gastrointestinal Endocopy and the American College of Gastroenterology [1, 5], are lower than ADRs presented elsewhere. We believe that the two main factors contributed our observed lower ADR: the high number of years in practice and low median annual volume of the practicing endoscopists. Both of these factors were associated with lower ADR in this study. This may limit the generalizability of our results and future research should determine whether our findings are reproducible in other centres.

For future investigations, we also recommend the analysis of advanced technologies such as EndoCuff, EndoRings, G-EYE™, FUSE, and cap-assisted colonoscopy. These techniques have been related to ADR rates of up 69%. It has been proposed that using these devices to flatten folds and improving endoscope stability can lead to narrowing the adenoma detection differences between the distal and proximal colon [32], however these technologies may be more applicable to augmenting ADR is specific areas of the colon. It has been previously demonstrated that feedback regarding overall ADR results in improvement in proximal but not distal adenoma detection [33]. An additional question of interest would be whether colon segment-specific performance feedback to colonoscopists can produce a more even distribution of improvement.

We found notable variability in endoscopists’ distal and proximal adenoma detection. While distal ADR and proximal ADR are both associated with increased overall ADR, endoscopist distal ADR was only moderately correlated with their proximal ADR. Information regarding both distal ADR and proximal ADR may complement currently used quality indicators to provide endoscopists with informative feedback regarding their adenoma detection proficiency.