External validation of four Pancreatic Fistula Risk Score models in the Deep South US: Do racial disparities affect pancreatic fistula prediction?

Conrad K. Blunck, Selwyn M. Vickers, Thomas N. Wang, Vikas Dudeja, Sushanth Reddy, J. Bart Rose

Research output: Contribution to journalArticlepeer-review

Abstract

Background: Fistula Risk Score (FRS) models often lack adequate discrimination for clinically relevant postoperative pancreatic fistula (CR-POPF) on external validation. We tested four FRS models in the Deep South United States and sought to determine if CR-POPF discrimination was affected by racial disparities. Methods: A single-institution retrospective cohort study of patients who underwent pancreatoduodenectomies between 2013 and 2019 was performed. FRS discrimination for CR-POPF was assessed using ROC curves for both the entire patient population, and for Black vs White patients. Results: The Alternative FRS maintains adequate CR-POPF discrimination when considering the patient population as a whole, but inadequately predicts CR-POPF when applied to the Black patient population. The Sun-FRS provides adequate CR-POPF discrimination for Black patients when considering risk grade. Only soft pancreatic gland texture and small duct size were significantly associated with CR-POPF in this patient population. Discussion: Institutions should assess their preferred FRS model to determine if it provides adequate CR-POPF discrimination among a racially diverse patient population. Further studies are needed to determine how racial disparities influence CR-POPF prediction to better guide postoperative management.

Original languageEnglish (US)
JournalAmerican journal of surgery
DOIs
StateAccepted/In press - 2022
Externally publishedYes

ASJC Scopus subject areas

  • Surgery

Fingerprint

Dive into the research topics of 'External validation of four Pancreatic Fistula Risk Score models in the Deep South US: Do racial disparities affect pancreatic fistula prediction?'. Together they form a unique fingerprint.

Cite this