Endoscopy 2020; 52(01): 15-16
DOI: 10.1055/a-1058-8854
Editorial
© Georg Thieme Verlag KG Stuttgart · New York

Detection rates during colonoscopy: What matters most?

Referring to Klair JS et al. p. 61–67
Michael Bretthauer
1  Department of Health Management and Health Economics, Institute of Health and Society, University of Oslo, Oslo, Norway
2  Department of Transplantation Medicine, Oslo University Hospital, Oslo, Norway
› Author Affiliations
Further Information

Publication History

Publication Date:
18 December 2019 (online)

The adenoma detection rate (ADR) is defined as the percentage of patients undergoing colonoscopy who have one or more adenomas detected [1] [2]. ADR has been the hallmark of high quality colonoscopy since a landmark study from Poland published in the New England Journal of Medicine in 2010 established the relationship between an individual endoscopist’s ADR and their patient’s risk of colorectal cancer (CRC) after colonoscopy (so-called interval cancer) [1]. The association of endoscopist’s ADR with CRC was confirmed by a large US cohort in 2014, and extended by establishing its importance also for CRC mortality [2].

“…it may be premature to establish a benchmark for serrated polyp detection (SDR) and advanced adenoma detection (AADR) rates … until the magnitude and pattern of the association of SDR and AADR with interval cancer risk, and their relationship to ADR, are established.”

Today, colonoscopy ADR is the most established quality indicator in gastrointestinal endoscopy. It is recognized as the key performance indicator in multiple guidelines, and its measurement and benchmarking is strongly recommended throughout the World [3] [4]. However, although there is no doubt that high ADR is strongly associated with desirable outcomes after colonoscopy, it is important to recognize that it is not a perfect or universal quality indicator.

1. ADR has been developed in colonoscopies that are performed for detection of colorectal adenomas (screening or surveillance). It may be difficult to apply to clinical colonoscopies with a patient population where polyp detection is not the primary focus of the intervention, such as in inflammatory bowel disease [3].

2. ADR is dependent on the prevalence of adenomas in the population. This prevalence is different in different countries and regions, and is dependent on patient age and sex [1] [2].

3. ADR may be prone to gaming by the endoscopist [5]. Endoscopists may be inclined to focus on finding one small adenoma in as many patients as possible, and have less interest in finding more than one because the number of adenomas is not incentivized when using ADR as a quality indicator.

4. ADR is dependent not only on endoscopist performance but also on the quality of the histopathology service.

5. Polyps other than adenomas may be independently related to interval CRC risk, such as serrated polyps or advanced adenomas [3].

6. Complete removal of detected adenomas may be more important than detection of adenomas and may differ between endoscopists with similar ADRs [6].

Because of these shortcomings, there is continued interest in exploring alternative quality indicators for colonoscopy performance related to the protection against CRC.

In this issue of Endoscopy, Klair et al. report on a multicenter cohort study at five medical centers in the United States to investigate whether detection rates for serrated polyps (SDR) or advanced adenomas (AADR), respectively, may be alternatives to ADR [7]. In accordance with current guidance, the authors restricted the serrated polyps that were used to calculate the SDR to serrated polyp ≥ 1 cm located anywhere in the colon or hyperplastic polyp ≥ 5 mm located proximally to the sigmoid colon (so-called clinically significant serrated polyps) [7].

The findings of a moderate association between ADR and SDR or AADR illustrated by a Spearman’s R value of 0.67 and 0.56, respectively is in line with previous studies, which have also found positive but not strong associations of ADR with SDR and AADR, respectively [8] [9] [10] [11].

The strengths of the new study are the consecutive sampling and the rigorous manual chart review, the inclusion of patients with a first colonoscopy, and the restriction to screening indication. Of note, almost all colonoscopists in the Klair et al. study had high ADRs (≥ 25 %), and colonoscopies with suboptimal bowel preparation and incomplete colonoscopies were excluded. Thus, the findings relate to performance measures in high quality colonoscopy services only.

If SDR and AADR were strongly correlated to ADR, it would mean that they measure the same thing. We then would have the luxury to choose what detection rate to use for quality assurance. Our choice would depend on the pros and cons of each of them as outlined above; for example, the risk of gaming or the yield that determines sample sizes required for quality monitoring (rare outcomes such as AADR require larger sample sizes compared with more frequent outcomes such as ADR). If they are not correlated strongly but still predict interval cancer risk, it means that they measure different things that are related to the same outcome – interval cancer. In that case, we may consider to measure both (i. e. introduce an additional key performance indicator to ADR).

To reliably answer these questions, we need to take a step back and reliably establish the association of SDR or AADR with the real outcome of interest – interval CRC after colonoscopy. To my knowledge, there are no studies to date that have presented data for that most interesting outcome. Without such studies, it is difficult to establish a quantification of the association of SDR and AADR with interval CRC. Such a quantification, however, is an absolute prerequisite for key quality indicators in medicine. Benchmarking is not advisable without it, and can only be derived through rigorous studies with hard end points (CRC incidence and mortality), as has been done so nicely for ADR [1] [2]. Consequently, it may be premature to establish a benchmark for SDR and AADR, as Klair et al. and others have tried [7] [11], until the magnitude and pattern of the association of SDR and AADR with interval cancer risk, and their relationship to ADR, are established.

There may be good reasons to move from ADR to SDR or AADR as the key quality indicator in colonoscopy. However, my recommendation would be not to establish SDR or AADR as quality indicators for individual endoscopist performance until we have evidence that is equally convincing as that for ADR, using the same end points (CRC incidence or mortality) and being of equally high research quality.