BACKGROUND: Recent studies on postoperative pancreatic fistula (POPF) prevention suggest that omission of perioperative drains is safe for negligible- or low-risk patients undergoing pancreatoduodenectomy (PD). However, this proposed pathway has not been validated in a nationwide cohort. STUDY DESIGN: The ACS-NSQIP-targeted pancreatectomy database from 2014 to 2016 was queried to identify patients who underwent PD. Using a previously validated modified Fistula Risk Score (mFRS), patients were stratified as negligible/low- or intermediate/high-risk. Multivariate regression models were used to analyze the effect of intraoperative drain placement on relevant perioperative outcomes in both high- and low-risk patients. RESULTS: Among 6,730 patients undergoing PD, 3,375 (50%) were high-risk; 3,355 (50%) were low-risk. Among high-risk patients, drain placement (n = 3,093, 92%) was associated with a higher rate of POPF (26% vs 16%, p = 0.0003), clinically relevant (CR) POPF (20% vs 12%, p = 0.0015), and extended hospital length of stay (LOS, 9 vs 7 days, p < 0.0001), but decreased serious morbidity (29% vs 35%, p = 0.0330). Similarly, drain placement in low-risk patients (n = 2,785, 83%) was associated with a higher rate of POPF (11% vs 6%, p = 0.0006) and extended LOS (8 vs 7 days, p < 0.0001), yet lower serious morbidity (18% vs 23%, p = 0.0037). On multivariate logistic regression, drain placement was associated with significantly increased odds of CR-POPF and a significantly reduced incidence of serious morbidity among both high-risk (odds ratio [OR] 0.72, 95% CI 0.55 to 0.94, p = 0.0155) and low-risk patients (OR 0.71, 95% CI 0.57 to 0.89, p = 0.0027). CONCLUSIONS: In this population-based cohort, the mFRS was unable to stratify patients relative to the need for selective drain placement during PD. For both high- and low-risk patients, perioperative drain placement was associated with increased rates of POPF, CR-POPF, and extended LOS, but decreased incidence of serious morbidity.
BACKGROUND: Recent studies on postoperative pancreatic fistula (POPF) prevention suggest that omission of perioperative drains is safe for negligible- or low-risk patients undergoing pancreatoduodenectomy (PD). However, this proposed pathway has not been validated in a nationwide cohort. STUDY DESIGN: The ACS-NSQIP-targeted pancreatectomy database from 2014 to 2016 was queried to identify patients who underwent PD. Using a previously validated modified Fistula Risk Score (mFRS), patients were stratified as negligible/low- or intermediate/high-risk. Multivariate regression models were used to analyze the effect of intraoperative drain placement on relevant perioperative outcomes in both high- and low-risk patients. RESULTS: Among 6,730 patients undergoing PD, 3,375 (50%) were high-risk; 3,355 (50%) were low-risk. Among high-risk patients, drain placement (n = 3,093, 92%) was associated with a higher rate of POPF (26% vs 16%, p = 0.0003), clinically relevant (CR) POPF (20% vs 12%, p = 0.0015), and extended hospital length of stay (LOS, 9 vs 7 days, p < 0.0001), but decreased serious morbidity (29% vs 35%, p = 0.0330). Similarly, drain placement in low-risk patients (n = 2,785, 83%) was associated with a higher rate of POPF (11% vs 6%, p = 0.0006) and extended LOS (8 vs 7 days, p < 0.0001), yet lower serious morbidity (18% vs 23%, p = 0.0037). On multivariate logistic regression, drain placement was associated with significantly increased odds of CR-POPF and a significantly reduced incidence of serious morbidity among both high-risk (odds ratio [OR] 0.72, 95% CI 0.55 to 0.94, p = 0.0155) and low-risk patients (OR 0.71, 95% CI 0.57 to 0.89, p = 0.0027). CONCLUSIONS: In this population-based cohort, the mFRS was unable to stratify patients relative to the need for selective drain placement during PD. For both high- and low-risk patients, perioperative drain placement was associated with increased rates of POPF, CR-POPF, and extended LOS, but decreased incidence of serious morbidity.