Using NSQIP data to determine when drains can be safely remove after pancreaticoduodenectomy

Purpose: Prior retrospective studies demonstrate the need for drainage after pancreatoduodenectomy. Early drain removal is a key component of most ERAS pathways; however, little consensus on timing and cutoff for amylase analysis exists. This study aimed to identify the ideal combination of postoperative day and drain amylase level for safe drain removal. Methods: : This retrospective cohort study used the NSQIP database to study outcomes of pancreatoduodenectomy patients. Results: : 7,583 patients underwent pancreatoduodenectomy with drain placement. 1,458 (19%) had a clinically relavent postoperative pancreatic fistula (CR-POPF). Of those, 7% had their highest drain amylase on day 1, compared to 29% and 64% on day 3-5 and 5, respectively (p 300 U/L corresponded to a 46% chance of CR-POPF compared to 4.9% with amylase <300 U/L (p 5000 U/L, corresponded to 71.2% chance of CR-POPF compared to 11.2% with amylase <5000 U/L (p 5000 U/L, an amylase cutoff of >300 U/L had a lower specificity on day 1 (93%, CI: 0.92, 0.93) and further decreases on postoperative day 3-5 (81%, CI: 0.79-0.98). The sensitivity of a cutoff >300 U/L was superior: 24% (CI: 0.19-0.29) on postoperative day 1 but increased significantly to 72% (CI: 0.67-0.76) on postoperative days 3-5. Conclusions: : Current recommendations utilizing 5000 U/L will miss 6% of patients with CR-POPF. We recommend checking drain amylase on postoperative day three and to use 300 U/L as a cutoff for early drain removal protocols.