Identification of significant difficulty of selective deep cannulation by a simple predictive model: an endoscopic scale for teaching ERCP.

BACKGROUND

Endoscopic retrograde cholangiopancreatography (ERCP) is an advanced endoscopic procedure associated with a risk of serious complications. This cohort study was designed to assess the accuracy of an endoscopic method aimed at discriminating procedures eligible to teach ERCP: patients with and without significant difficulty of selective deep cannulation (DSDC).

METHODS

Clinically relevant variables were analyzed in a cohort of 400 consecutive patients (estimation group = 250 patients; validation group = 150 patients) who underwent an ERCP procedure.

RESULTS

Multivariate analysis identified fixated duodenum, inflamed duodenum, soft major papilla, previous biliary surgery, and papilla with ectopion as independent predictors of DSDC. We constructed a model and a score system combining these five variables. The area under the ROC curve was 0.81 for the estimation group and 0.80 for the validation group. Using the best cutoff score (> 1.63), absence of significant DSDC could be excluded with high accuracy (negative predictive value = 89.2%) in 111 (44.4%) of 250 patients. Similarly, it could be excluded with the same certainty in 77 (51.3%) of the 150 patients in the validation group.

CONCLUSIONS

A combination of easily accessible variables accurately predicts the absence of significant DSDC in half the patients who underwent the ERCP procedure. This score system discriminates procedures eligible to teach ERCP.