Subscribe to RSS
DOI: 10.1055/s-0039-1681206
PREDICTING ADVERSE CLINICAL OUTCOMES FOR COLORECTAL ENDOSCOPIC SUBMUCOSAL DISSECTION (CR-ESD): SMSA VERSUS A NEW EXPERIENCE-LESION SCORE. A MULTICENTER SPANISH PROSPECTIVE STUDY
Publication History
Publication Date:
18 March 2019 (online)
Aims:
To develop a new preoperative model to improve the ability of the SMSA score to predict adverse clinical outcomes of CR-ESD: duration of the procedure > 240 min., piecemeal resections, aborted procedures and complications (intraprocedural and delayed perforations and delayed bleedings).
Methods:
Consecutive patients were enrolled in a prospective multicenter Spanish CR-ESD registry since January 2016 to October 2018. We analyzed 585 cases in 19 hospitals. The overall ability of both scores to discriminate between those who developed adverse outcomes and those who did not was assessed by the area under the ROC curve.
Results:
Overall, 221 cases developed any of the predefined adverse outcomes. The AUROC of the SMSA score >= 3 was 0.51 (CI95%: 0.46 – 0.55). Thus, an alternative logistic regression model was designed. It included significant variables that were associated with the predefined outcomes in the univariate analysis. One of them was related with the experience of the endoscopic team, case load ≤10 lesions: OR = 4.5 (CI95%: 1.5 – 13.2; p = 0.007) and the remaining were associated with characteristics of the lesion: poor manoeuvrability, OR = 1.6 (CI95%: 1.1 – 2.2; p = 0.007), size > 30 mm, OR = 1.5 (CI95%: 1.01 – 2.2; p = 0.02), LST-G mixed type with a nodule > 10 mm, OR = 2.8 (CI95%: 1.1 – 7.1; p = 0.03) and previous endoscopic electrosurgical treatment, OR = 2.2 (CI95%: 1.06 – 4.6; p = 0.03). The AUROC for this multivariate model was 0.61 (CI95%: 0.57 – 0.66). The difference between both AUROCs was statistically significant (p < 0.00001).
Conclusions:
The SMSA score was useless to predict adverse outcomes for CR-ESD. A new score based on a multivariate logistic regression model, the Experience-Lesion score, showed better discrimination abilities to predict these unfavourable events.