Development of clinical prediction rule for the requirement of endoscopic papillary large balloon dilation (EPLBD) on endoscopic CBD stone clearance

IntroductionTo develop a simplified scoring system for clinical prediction of difficulty in CBD stone removal to assist endoscopists working in resource-limited settings in deciding whether to proceed with an intervention or refer patients to a center capable of performing additional procedures and interventions. MethodsThis study included patients with CBD stones who underwent ERCP at Pattani Hospital between August 2017 and December 2021. Retrospective cohort data was collected and patients were categorized into two groups: bile duct stones successfully treated by endoscopic biliary sphincterotomy and extraction compared to the former method combined with EPLBD. We explored potential predictors using multivariable logistic regression. The chosen logistic coefficients were transformed into a scoring system based on risk with internal validation via bootstrapping procedure. ResultsAmong the 155 patients who had successful endoscopic therapy for bile duct stones, there were 79 (50.97%) cases of endoscopic biliary sphincterotomy, EPLBD and extraction versus 76 (49.03%) cases without EPLBD. The factors used to derive a scoring system included the size of CBD stones >15 mm, the difference between the stone and distal CBD diameter >2mm, distal CBD arm length <36 mm and stone shape. The score-based model's area under ROC was 0.88 (95% CI: 0.83, 0.93). For clinical use, the range of scores from 0 to 16, was divided into two subcategories based on CBD stone removal difficulty requiring EPLBD to derive the PPV. For scores <5 and >= 5, the PPV was 23.40 (p <0.001) and 93.44 (p <0.001) respectively. The Bootstrap sampling method indicated a prediction ability of 0.88 (AuROC, 95% CI: 0.83, 0.94). ConclusionThis scoring system has acceptable prediction performance in assisting endoscopists in their choice of stone removal procedure.

Automated summary

A simplified scoring system was developed to predict the difficulty of CBD stone removal in clinical practice, which can assist endoscopists in making treatment decisions. Retrospective cohort data was collected and analyzed using multivariable logistic regression. The scoring system based on risk factors showed good prediction performance and was internally validated.