This media is currently not available.
Predicting Post‑Clearance Stone Events After ERCP: A Simple Real‑Life Risk Model
Poster Abstract

Aims

Despite technically successful ERCP, a subset of patients re-presents with stone-related events requiring repeat interventions. Current ESGE guidance acknowledges recurrence but offers no practical strategy to identify high‑risk individuals. This study aimed to determine real‑life predictors of post‑clearance events and develop a simple, immediately applicable risk model.

Methods

Consecutive patients with confirmed choledocholithiasis undergoing therapeutic ERCP were retrospectively analyzed. Only procedures with strict verification of complete clearance (stone extraction and clean cholangiogram) were included. Stone‑related reintervention was defined as any repeat ERCP for residual or recurrent stones. Clinical, radiologic, and endoscopic variables were assessed. A concise 3‑item risk score was constructed based on the strongest predictors.

Results

Among 494 patients with complete duct clearance, 33 (6.7%) required stone‑related reintervention. Stone burden was the dominant determinant: empierrement (≥3 stones) increased risk nearly fourfold (12.4% vs 3.7%; OR 3.7). Papillary anatomical difficulty (bombante, deformed, or peri‑diverticular papilla) showed a synergistic impact. In contrast, stone size ≥15 mm, CBD dilation ≥12 mm, and prior cholecystectomy were not independent predictors. The resulting 3‑item Stone‑Risk Score (empierrement + papillary difficulty + pre‑ERCP drainage) stratified risk from 2.1% (score 0) to 14.8% (score ≥2), with meaningful discrimination (AUC ≈ 0.72).

Conclusions

Post‑clearance stone events are predictable rather than random. Stone burden and papillary complexity—rather than stone size or duct dilation—drive recurrence risk. This pragmatic model provides immediate clinical value and can refine post‑ERCP follow‑up strategies. These findings highlight a gap in current ESGE recommendations and justify targeted surveillance for high‑risk profiles.