Intended for healthcare professionals

  1. Education
  2. Computer aided...
  3. Computer aided detection and diagnosis of polyps in adult patients undergoing colonoscopy: a living clinical practice guideline
Practice Rapid Recommendations

Computer aided detection and diagnosis of polyps in adult patients undergoing colonoscopy: a living clinical practice guideline

BMJ 2025; 388 doi: https://doi.org/10.1136/bmj-2024-082656 (Published 27 March 2025) Cite this as: BMJ 2025;388:e082656
Visual summary of recommendation
Population People undergoing colonoscopy for any indication Recommendation applies to: Recommendation does not apply to: Individuals undergoing colonoscopy for: Screening Follow up of positive faecal immunochemical testing Surveillance Adult patients (18 years and older) who have agreed to undergo a colonoscopy a history of inflammatory bowel disease abnormal imaging findings therapeutic interventions
Recommendation 1
Routine colonoscopyEndoscopic examination of the entire colonCADe colonoscopyComputer aided detection (CADe)colonoscopy - supplemented with use ofartificial intelligence to increase adenomadetection rateorWe suggest against the routine use of CADe colonoscopyStrongAll or nearly all informed people would likelywant the intervention to the left. Benefitswould outweigh harms for almost everyoneWeakMost people would likely want the interventionto the left. Benefits would outweigh harms forthe majority, but not for everyoneWeakMost people would likely want the interventionto the right. Benefits would outweigh harmsfor the majority, but not for everyoneStrongAll or nearly all informed people would likelywant the intervention to the right. Benefitswould outweigh harms for almost everyone

potential benefits

Favours routine colonoscopyFavours CADe colonoscopyNo important differenceThe panel found that this difference was not important for most patients,because the intervention effects were negligible and/or very imprecise, forexample confidence intervals that include both important benefit and harm
Within 10 yearsper 10 000 peopleEvidence quality

Colorectal cancer incidenceNo important difference8271Very lowMore

Based on microsimulation study - no trial data availableWe are uncertain whether CADecolonoscopy improves or worsencolorectal cancer incidenceVery lowGRADE rating,because of:Risk of biasSeriousImprecisionSeriousIndirectnessNo serious concernsInconsistencyNo serious concernsPublication biasNo serious concernsGRADE certainty ratingsHighThe authors have a lot ofconfidence that the trueeffect is similar to theestimated effectModerateThe authors believe thatthe true effect is probablyclose to theestimated effectLowThe true effect might bemarkedly different fromthe estimated effectVery lowThe true effect isprobably markedlydifferent from theestimated effect

Colorectal cancer related deathsNo important difference1513LowMore

Based on microsimulation study - no trial data availableCADe colonoscopy may have little or nodifference on colorectal cancer relateddeathsLowGRADE rating,because of:Risk of biasNo serious concernsImprecisionSeriousIndirectnessNo serious concernsInconsistencyNo serious concernsPublication biasNo serious concernsGRADE certainty ratingsHighThe authors have a lot ofconfidence that the trueeffect is similar to theestimated effectModerateThe authors believe thatthe true effect is probablyclose to theestimated effectLowThe true effect might bemarkedly different fromthe estimated effectVery lowThe true effect isprobably markedlydifferent from theestimated effect

Colorectal cancer incidenceNo important difference3423Very lowMore

Based on microsimulation study - no trial data availableWe are uncertain whether CADecolonoscopy improves or worsenpost-colonoscopy colorectal cancerincidenceVery lowGRADE rating,because of:Risk of biasSeriousImprecisionSeriousIndirectnessNo serious concernsInconsistencyNo serious concernsPublication biasNo serious concernsGRADE certainty ratingsHighThe authors have a lot ofconfidence that the trueeffect is similar to theestimated effectModerateThe authors believe thatthe true effect is probablyclose to theestimated effectLowThe true effect might bemarkedly different fromthe estimated effectVery lowThe true effect isprobably markedlydifferent from theestimated effect

Surveillance colonoscopy635 fewer26453280LowMore

Based on microsimulation study - no trial data availableCADe colonoscopy may increase number ofcolonoscopies per lifetimeLowGRADE rating,because of:Risk of biasSeriousImprecisionNo serious concernsIndirectnessNo serious concernsInconsistencyNo serious concernsPublication biasNo serious concernsGRADE certainty ratingsHighThe authors have a lot ofconfidence that the trueeffect is similar to theestimated effectModerateThe authors believe thatthe true effect is probablyclose to theestimated effectLowThe true effect might bemarkedly different fromthe estimated effectVery lowThe true effect isprobably markedlydifferent from theestimated effect
Intraoperativelyper 100 peopleEvidence quality

Adenoma detection rate8 more3745LowMore

Based on data from 30 674 patients in 40 studiesCADe colonoscopy may improve adenomadetection rateLowGRADE rating,because of:Risk of biasSeriousImprecisionNo serious concernsIndirectnessNo serious concernsInconsistencyNo serious concernsPublication biasSeriousGRADE certainty ratingsHighThe authors have a lot ofconfidence that the trueeffect is similar to theestimated effectModerateThe authors believe thatthe true effect is probablyclose to theestimated effectLowThe true effect might bemarkedly different fromthe estimated effectVery lowThe true effect isprobably markedlydifferent from theestimated effect
See all outcomes
Patient decision aids

Rationale

Benefits of CADe colonoscopy on critical outcomes, such as colorectalcancer incidence and post-colonoscopy cancer incidence, are highlyuncertain. Low certainty evidence suggests little to no impact oncolorectal cancer-related mortality, while the potentialburdens—including more frequent surveillance colonoscopies, increasedhealth-related anxiety, and overdiagnosis—are likely to affect manypatients. Given the trivial and uncertain benefits and the likelihoodof burdens, the panel issued a weak recommendation against routineCADe colonoscopy use.

Individual considerations

Key practical issuesCADe colonoscopyWhen used, CADe colonoscopy should not be a substitute for a careful andthorough examinationAny potential benefit of CADe colonoscopy is contingent upon colonoscopybeing performed by a trained endoscopist who ensures full mucosalvisualisation of the colonDespite speculations that CADe colonoscopy is most beneficial for noviceendoscopists, there is no evidence to support conclusions on its efficacybeing modified by the endoscopist’s skill level
Values and preferences• Given uncertain benefits on critical outcomes and potential burdens (morefrequent surveillance colonoscopies, increased health-related anxiety,overdiagnosis), most well-informed patients who have decided to undergocolonoscopy may not favour CADe assistance.• Patients may place more value on avoiding potential burdens than on currentlyuncertain benefits concerning critical outcomes such as colorectal cancerincidence, cancer-related mortality, and post-colonoscopy cancer incidence.• Limited evidence exists regarding patient values and preferences regardingCADe colonoscopy use. The weak recommendation against use was informed by60% of panel members (including patient partners) voting for a recommendationagainst CADe use and 40% voting in favour.
Population and system-level considerations• Decision analytic modelling shows that the implementation of CADe may resultin a substantial increase in follow-up visits and procedures.• These, in turn, may result in a substantial increase in costs and a reduction inavailable resources.• The criteria for approval of CADe devices from regulatory agencies such as theUS Food and Drug Administration remains unclear, making it difficult to integrateand compare different CADe systems in practice.

Additional areas of uncertainty

The impact of CADe colonoscopy on colorectal cancer incidence andrelated mortality is based on modelling work with low to very lowcertainty.Most currently available randomised trials evaluating CADe colonoscopyare unblinded, raising concerns regarding provider bias and possiblyinfluencing outcomes (overestimating benefits).

©BMJ Publishing Group Limited.

Disclaimer: This infographic is not a validated clinical decision aid. This information is provided without any representations, conditions or warranties that it is accurate or up to date. BMJ and its licensors assume no responsibility for any aspect of treatment administered with the aid of this information. Any reliance placed on this information is strictly at the user's own risk. For the full disclaimer wording see BMJ's terms and conditions: https://www.bmj.com/company/legal-information/

Find recommendations, evidence summaries and consultation decision aids for use in your practice

BMJ Medicine Research

Benefits, burden, and harms of computer aided polyp detection with artificial intelligence in colorectal cancer screening: microsimulation modelling study
Author Information

Please note:
  • Rapid responses are electronic comments to the editor. They are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles.
  • Your name will be published with your response. Include your email address in the text of your response if you want others to see it.
  • Once posted, you will not have the right to remove or edit your response. The BMJ may remove or edit responses at its absolute discretion, including when it is brought to our attention that a response spreads misinformation.
  • A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed
  • Rapid responses have their own URL and are retrievable in an advanced search of thebmj.com in perpetuity.
  • If patients could recognise themselves, or anyone else could recognise a patient from your description, please obtain the patient's written consent to publication before sending your response. See our patient consent form.
  • The word limit for rapid responses is 600 words (excluding references).
  • By submitting this rapid response your are agreeing to our full rapid response requirements.
  • Please do not include original data in your response, unless it has already been published in a peer reviewed journal and you are able to include a reference.

Note: this will be visible to readers on the site. Please only include information you are comfortable with being published.

Statement of Competing Interests

A competing interest exists when professional judgment concerning a primary interest (such as patients' welfare or the validity of research) may be influenced by a secondary interest (such as financial gain or personal rivalry). Find out more.

Compose response

Note: avoid using generic titles like 'in response to the authors' - be specific.

More information about text formats

Plain text

  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Terms and Conditions

Read terms and conditions

Vertical Tabs

Back to top