Surgical Drains Are Associated with Decreased Severity of Pancreatic Fistulas in Pancreaticoduodenectomy. an Analysis of the Nsqip Pancreatectomy Database

Background: Despite multiple randomized controlled trials controversy continues regarding the effectiveness of surgical drains in pancreaticoduodenectomy. We hypothesized that while drains may not affect the frequency of pancreatic fistulas, they might reduce the frequency of severe complications owing to earlier detection and source control. To this end, we queried the procedure targeted Pancreatectomy NSQIP database to determine the effect of post-operative drains on re-intervention, and severe fistulas post pancreaticoduodenectomy. Methods: We performed a retrospective cohort study of the procedure targeted, Pancreatectomy NSQIP database from 2014-2016. We included all patient who underwent non-emergent, pancreaticoduodenectomy with pancreaticojejunostomy or pancreaticogastrostomy. We excluded patients who had sepsis at the time of OR, or who underwent emergent pancreaticoduodenectomy. Primary endpoint was a modified Clinically relevant fistula, which included any fistula that required either percutaneous drainage, or was associated with sepsis, organ failure, reoperation, or death. Secondary endpoints evaluated were percutaneous drainage, sepsis, septic shock, acute renal failure, transfusion after pod 1, re-operation, cardiac arrest, organ space infection, and 30-day mortality. Multivariable logistic regression models were generated for the primary outcome adjusting for known confounders, and individually for each secondary outcome. Results: Between 2014 and 2016, 10 094 pancreaticoduodenectomies were performed that met our inclusion criteria. Post-operative drains were left in 91.5% of pancreaticoduodenectomies and omitted in 8.5%. On univariable logistic regression, drain placement was not associated with modified CR fistula. After adjustment, post-operative drain placement was associated with a 33.6% reduction in the odds of modified CR fistula (OR 0.674, 95% CI 0.538-0.844, P = 0.0006). When we evaluated the effect of post-operative drain placement on secondary end points (Table 1), drains were associated with fewer percutaneous drains (Hard pancreas: OR 0.82, 95%CI 0.471-0.980, p= 0.0387, Soft Pancreas: OR 0.488, 95% CI 0.370–0.644, p < 0.0001), septic shock (OR 0.555 95%CI 0.396 – 0.776, p = 0.0006), renal failure (Soft pancreas: OR 0.246, 95% CI 0.109-0.559, p= 0.0008), cardiac arrest (OR 0.452, 95% CI 0.285 – 0.717, p = 0.0007), and organ space infection (OR 0.657, 95%CI 0.550 – 0785, p = <0.0001). Conclusion: Based on our analysis of the NSQIP procedure targeted pancreatectomy database, and 10 094 pancreaticoduodenectomies, post-operative drains are associated with decreased odds of clinically relevant endpoints including modified CR fistulas, percutaneous drainage, septic shock, cardiac arrest and organ space infection. These associations were also noted even in patients with a firm pancreatic texture. Based on this data it may be prudent to continue post-operative drain placement, and further investigations should focus on these important outcomes in addition to the ISGPS definitions of CR pancreatic fistulas.