In:
Endoscopy International Open, Georg Thieme Verlag KG, Vol. 11, No. 08 ( 2023-08), p. E724-E732
Abstract:
Background and study aims Overcoming logistical obstacles for the implementation of colorectal endoscopic submucosal dissection (ESD) requires accurate prediction of procedure times. We aimed to evaluate existing and new prediction models for ESD duration. Patients and methods Records of all consecutive patients who underwent single, non-hybrid colorectal ESDs before 2020 at three Dutch centers were reviewed. The performance of an Eastern prediction model [GIE 2021;94(1):133–144] was assessed in the Dutch cohort. A prediction model for procedure duration was built using multivariable linear regression. The model’s performance was validated using internal validation by bootstrap resampling, internal-external cross-validation and external validation in an independent Swedish ESD cohort. Results A total of 435 colorectal ESDs were analyzed (92% en bloc
resections, mean duration 139 minutes, mean tumor size 39 mm). The performance of current unstandardized time scheduling practice was suboptimal (explained variance: R2=27%). We
successfully validated the Eastern prediction model for colorectal ESD duration 〈 60 minutes
(c-statistic 0.70, 95% CI 0.62–0.77), but this model was limited due to dichotomization of the outcome and a relatively low frequency (14%) of ESDs completed 〈 60 minutes in the Dutch
centers. The model was more useful with a dichotomization cut-off of 120 minutes (c-statistic: 0.75; 88% and 17% of “easy” and “very difficult” ESDs completed 〈 120 minutes,
respectively). To predict ESD duration as continuous outcome, we developed and validated the six-variable cESD-TIME formula (https://cesdtimeformula.shinyapps.io/calculator/; optimism-corrected
R2=61%; R2=66% after recalibration of the slope). Conclusions We provided two useful tools for predicting colorectal ESD duration at Western centers. Further improvements and validations are encouraged with potential local adaptation to optimize time planning.
Type of Medium:
Online Resource
ISSN:
2364-3722
,
2196-9736
Language:
English
Publisher:
Georg Thieme Verlag KG
Publication Date:
2023
detail.hit.zdb_id:
2761052-4
