Clinical Prediction Models to Guide the Selection of Patients for Colon Capsule Endoscopy Versus Colonoscopy...

Area of Science:

• Gastroenterology
• Medical Informatics
• Clinical Decision Making

Background:

• Colonoscopy demand strains healthcare systems globally.
• Colon capsule endoscopy (CCE) is a less invasive alternative but faces challenges with high re-investigation rates.
• Effective patient selection is crucial for optimizing CCE use after positive fecal immunochemical tests (FIT).

Purpose of the Study:

• To develop and evaluate clinical prediction models for selecting FIT-positive patients for CCE versus colonoscopy.
• To assess the models' ability to predict CCE outcomes and colonoscopy necessity.
• To determine the clinical utility of these models in guiding post-FIT triage decisions.

Main Methods:

• Secondary analysis of the CareForColon2015 randomized controlled trial data (2020-2022).
• Development of logistic regression models using 60 candidate predictors to forecast CCE transit, cleansing, completeness, and colonoscopy indication.
• Model validation via repeated random subsampling and evaluation on a hold-out set using AUC, Cohen's K, and accuracy; Decision Curve Analysis (DCA) for clinical utility.

Main Results:

• CCE achieved complete transit in 92.1% and acceptable cleansing in 71.3%; 69.6% of investigations were deemed complete.
• Colonoscopy was indicated in 68.0% (broad criteria) and 55.9% (stringent criteria).
• Prediction models for colonoscopy indication showed moderate performance (AUC 0.69-0.71). DCA indicated potential net benefit for identifying patients unlikely to need immediate colonoscopy.

Conclusions:

• Clinical prediction models show promise for guiding post-FIT triage between CCE and colonoscopy.
• These models may help reduce unnecessary colonoscopies by identifying suitable candidates for initial CCE.
• External validation is recommended prior to clinical implementation of these predictive tools.