Preoperative prediction of postoperative pancreatic fistula after Pancreaticoduodenectomy: Determination and validation of a cut-off value for the Roberts Score.

BACKGROUND

POPF after pancreaticoduodenectomy can be life-threatening. For risk stratification, prediction could be key. The aim of this study is to determine and validate a cut-off value for the Roberts Score, which is one of the few purely preoperative multicenter validated predictive models for POPF.

METHODS

582 patients were included. The Youden index determined a cut-off in the exploratory cohort (n ​= ​466). The validation cohort's (n ​= ​116) ability to predict CR-POPF was tested using univariate and multivariate regression analysis.

RESULTS

AUC of Roberts Score for the exploration cohort was 0.768. The identified cut-off of 0.268 was confirmed in the validation cohort (p ​< ​0.001). Higher scores were significantly associated with longer time to drain removal and ICU stay. Multiple logistic regression showed the cut-off as an independent predictor of CR-POPF (p ​= ​0.038).

CONCLUSION

The scoring variables and the cut-off itself were both independent predictors, which may improve the identification of high-risk patients and help to investigate the development of POPF.