RESUMO
BACKGROUND: Early drain removal when postoperative day (POD) one drain fluid amylase (DFA) was ≤5000 U/L reduced complications in a previous randomized controlled trial. We hypothesized that most surgeons continue to remove drains late and this is associated with inferior outcomes. METHODS: We assessed the practice of surgeons in a prospectively maintained pancreas surgery registry to determine the association between timing of drain removal with demographics, comorbidities, and complications. We selected patients with POD1 DFA ≤5000 U/L and excluded those without drains, and subjects without data on POD1 DFA or timing of drain removal. Early drain removal was defined as ≤ POD5. RESULTS: Two hundred and forty four patients met inclusion criteria. Only 90 (37%) had drains removed early. Estimated blood loss was greater in the late removal group (190 mL versus 100 mL, P = 0.005) and pathological findings associated with soft gland texture were more frequent (97 [63%] versus 35 [39%], P < 0.0001). Patients in the late drain removal group had more complications (84 [55%] versus 30 [33%], P = 0.001) including pancreatic fistula (55 [36%] versus 4 [4%], P < 0.0001), delayed gastric emptying (27 [18%] versus 3 [3%], P = 0.002), and longer length of stay (7 d versus 5 d, P < 0.0001). In subset analysis for procedure type, complications and pancreatic fistula remained significant for both pancreatoduodenectomy and distal pancreatectomy. CONCLUSIONS: Despite level one data suggesting improved outcomes with early removal when POD1 DFA is ≤ 5000 U/L, experienced pancreas surgeons more frequently removed drains late. This practice was associated with known risk factors (estimated blood loss, soft pancreas) and may be associated with inferior outcomes suggesting potential for improvement.