Extended Experience with a Dynamic, Data-Driven Selective Drain Management Protocol in Pancreaticoduodenectomy: Progressive Risk Stratification for Better Practice.

BACKGROUND: Intraoperative drain use for pancreaticoduodenectomy has been practiced in an unconditional, binary manner (placement/no placement). Alternatively, dynamic drain management has been introduced, incorporating the Fistula Risk Score (FRS) and drain fluid amylase (DFA) analysis, to mitigate clinically relevant postoperative pancreatic fistula (CR-POPF).

STUDY DESIGN: An extended experience with dynamic drain management was used at a single institution for 400 consecutive pancreaticoduodenectomies (2014 to 2019). This protocol consists of the following: drains omitted for negligible/low-risk FRS (0 to 2) and drains placed for moderate/high-risk FRS (3 to 10) with early (postoperative day [POD] 3) removal if POD1 DFA ≤5,000 U/L. Adherence to this protocol was prospectively annotated and outcomes were retrospectively analyzed.

RESULTS: The overall CR-POPF rate was 8.7%, with none occurring in the negligible/low-risk cases. Moderate/high-risk patients manifested an 11.9% CR-POPF rate (n = 35 of 293), which was lower on-protocol (9.5% vs 21%; p = 0.014). After drain placement, POD1 DFA ≥5,000 U/L was a better predictor of CR-POPF than FRS (odds ratio 14.7; 95% CI, 4.3 to 50.3). For POD1 DFA ≤5,000 U/L, early drain removal was associated with fewer CR-POPFs (2.8% vs 23.5%; p < 0.001), and substantiated by multivariable analysis (odds ratio 0.09; 95% CI, 0.03 to 0.28). Surgeon adherence was inversely related to CR-POPF rate (R = 0.846).

CONCLUSIONS: This extended experience validates a dynamic drain management protocol, providing a model for better drain management and individualized patient care after pancreaticoduodenectomy. This study confirms that drains can be safely omitted from negligible/low-risk patients, and moderate/high-risk patients benefit from early drain removal.