Endoscopy 2021; 53(05): 509-510
DOI: 10.1055/a-1290-7610

Doing our best to do no harm

Referring to Benazzato L et al. p. 501–508
Uri Ladabaum
Division of Gastroenterology and Hepatology, Stanford University School of Medicine, Stanford, California, United States
› Author Affiliations

First do no harm. Colorectal cancer (CRC) screening is an exemplary public health success – but the vast majority of people will never develop CRC, with or without screening. The great successes of CRC screening require exposing very large numbers of people to the potential risks of screening. The study by Benazzato et al. of adverse events post-colonoscopy in Veneto’s CRC screening program [1] makes a valuable contribution by focusing on a population undergoing colonoscopy primarily because of an abnormal fecal immunochemical test (FIT).

“The observation that the occurrence of adverse events was strongly associated with 30-day post-colonoscopy death makes a strong case that some post-colonoscopy deaths are indeed related to the procedure.”

The safety of colonoscopy has rightly been the focus of previous research and quality assurance guidance. Several systematic reviews and meta-analyses have pooled the published data. Most studies have focused on bleeding and perforation. There is much less information on serious nongastrointestinal events.

The data of Benazzato et al. [1] are in line with previous data: the bleeding and perforation rates are higher than those in screening populations [2], but very similar to the pooled estimates for a broader spectrum of indications [3] [4]. Individuals with a FIT + result represent a population enriched in advanced neoplasia [5], in whom a higher rate of polypectomy, larger polyps, and endoscopic mucosal resection (EMR) would be expected. The findings of Benazzato et al. confirm the previously identified risk factors for adverse events.

A systematic review performed for the United States Preventive Services Task Force that focused on screening populations reported major bleeding and perforation rates of 8 (95 % confidence interval [CI] 5 – 14) and 4 (95 %CI 2 – 5) per 10 000 procedures, respectively, with 96 % of major bleeds and 35 % of perforations occurring after polypectomy [2]. A second systematic review that considered a broader range of colonoscopy indications reported major bleeding and perforation rates of 26 (95 %CI 17 – 37) and 5 (95 %CI 4 – 7) per 10 000 procedures, respectively, with contrasts for polypectomy (98 [95 %CI 77 – 121] and 8 [95 %CI 6 – 10] per 10 000, respectively) vs. no polypectomy (6 [95 %CI 2 – 11] and 4 [95 %CI 2 – 8] per 10 000, respectively), and for procedures for symptoms vs. screening/surveillance [3]. A death rate of 2.9 (95 %CI 1.1 – 5.5) per 100 000 colonoscopies was reported. Finally, a systematic review performed for the American Society for Gastrointestinal Endoscopy reported major bleeding and perforation rates of 24 (95 %CI 24 – 25) and 5.8 (95 %CI 5.7 – 6.0) per 10 000 procedures, respectively, with a 2.7 % increase in bleeding risk for every 1 % increase in the polypectomy rate (P < 0.001) [4]. Deaths attributable to colonoscopy were estimated as 3 per 100 000 colonoscopies. With EMR of polyps ≥ 20 mm, the delayed bleeding and perforation rates were 400 (95 %CI 350 – 450) and 110 (95 %CI 90 – 140) per 10 000 procedures, respectively, with proximal location as a risk factor for delayed bleeding [4].

Most previous studies have focused on bleeding and perforation, and do not include a control group, leaving questions about the expected rates without colonoscopy. Our recent population-based study focused on severe post-colonoscopy gastrointestinal and nongastrointestinal adverse events, including as comparators three types of low-risk outpatient procedures [6]. After screening/surveillance colonoscopies, the rates of lower gastrointestinal bleeding, perforation, myocardial infarction, and ischemic stroke events per 10 000 persons were 5.3 (95 %CI 4.8 – 5.9), 2.9 (95 %CI 2.5 – 3.3), 2.5 (95 %CI 2.1 – 2.9), and 4.7 (95 %CI 4.1 – 5.2) without biopsy/intervention, and 36.4 (95 %CI 35.1 – 37.6), 6.3 (95 %CI 5.8 – 6.8), 4.2 (95 %CI 3.8 – 4.7), and 9.1 (95 %CI 8.5 – 9.7) with biopsy/intervention. Event rates were higher after non-screening/-surveillance examinations. The myocardial infarction, stroke, and serious pulmonary event rates were no higher than after low-risk comparator procedures. Myocardial infarction rates were similar to those published for the general US population.

One particularly thorny question is whether colonoscopy is associated with a risk of death. No study has been able to control adequately for the expected death rate in a comparable noncolonoscopy cohort. It is unclear whether the ongoing randomized controlled trials of screening colonoscopy will have the power to address this question. The observation by Benazzato et al. [1] that the occurrence of adverse events was strongly associated with 30-day post-colonoscopy death makes a strong case that some post-colonoscopy deaths are indeed related to the procedure. Of the 15 deaths reported, the cause of death was perforation in 3, myocardial ischemia in 7, cardiac arrest in 1, and other in 4. Our previous study [6] suggests that most post-colonoscopy myocardial infarctions may be part of the background population event rate, but it would seem highly coincidental for post-colonoscopy perforations leading to death not to be causally related to the colonoscopy.

There are lessons and lingering questions for colonoscopists in the clinical trenches, for directors of screening programs, and for researchers. It is not possible for clinicians to gain an intuitive feel for the magnitude of the risks of colonoscopy based on their own experience. We should counsel patients and practice based on published data reflecting very large denominators. Cold snare polypectomy has emerged as the resection method of choice for small polyps. Clipping should be performed after removal of proximal polyps ≥ 20 mm [7]. The observation by Benazzato et al. [1] that perforation was more common in procedures performed after incomplete colonoscopy should remind us of the common-sense advice [8] not to push against fixed resistance, to reduce the scope, and to continuously strive to improve our technique.

How to ascertain rare events reliably remains a challenge. This is nearly impossible outside of a dedicated research study or a fully integrated system. Many patients seek care for complications away from where a colonoscopy was performed. Thousands of post-procedure telephone calls would be needed before a handful of adverse events were uncovered – and failure to reach patients due precisely to post-procedure complications would result in substantial undercounting. One practical suggestion is to treat each identified “sentinel event” [8] as an opportunity for practice improvement.

Looking forward, electronic innovations to capture adverse events could aid practice improvement efforts. More data are needed on the CRC risk vs. the polypectomy risk for sessile serrated polyps. It is encouraging that post-colonoscopy bleeding risk appears to be declining [3]. As enthusiasts of CRC screening, let us continue to strive to optimize the benefit vs. harm balance of this very powerful public health endeavor.

Publication History

Publication Date:
22 April 2021 (online)

© 2021. Thieme. All rights reserved.

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany